High Level Game Design- Separating Characterisation from Plot

[HaldaneBDoyle]

I'm new to creating in RenPy but I have been a fan of interactive fiction (including select visual novels) for years. I started seriously writing novels a few years ago, but recently decided to shift focus to work on improving my skills at making visual novels instead (which has included a crash course on interactive fiction theory and writing for video games which got me even more excited).

One major problem with branching narratives is the exponential explosion of options over time. One possible solution is the "storylet" system- breaking the content into smaller discrete units (with a range of possible mechanics to integrate them into one overarching world or narrative).

For my game, I want to try something that I haven't seen done before (but please correct me if I have missed a comparable example).

The broad idea is to break the game into three distinct phases.

1. Establish the main character and premise as efficiently as possible. Give the player some stakes/choices with a linear narrative that only otherwise leads to terminal side branches (a classic gauntlet structure). The reasoning is that restarting a game after investing 5 minutes isn't too off-putting.

2. Establish the main secondary characters. This is the main innovation I want to figure out. The idea is to write a large pool of storylets that cover the entire backstory of the NPC. Rather than a small dialogue tree you can mine for all the possible options, the dialogue will be made of a large swarm of storylets (estimating about a hundred of them, each a few hundred words long). Each storylet will potentially link to a number of hand-crafted follow-up storylets (say 6 options each) but the player will only be presented a random subset in any playthrough (say 2 or 3, with no repeating storylets). The time available to converse with the NPC will be limited, so you can only cover ~20-30% in a single playthrough (increasing replayability if the player enjoys learning about the NPC, and creating an "iceberg" effect of unexplored backstory). During this phase the player accumulates stats/flags/etc that don't directly impact the unfolding conversation in this stage of the game.

3. Explore the plot via a compact branching narrative that drives the story to its climax. Leaving the standard branching narrative this late in the game means the exponential explosion can be kept manageable, while still allowing lots of options (which open/close depending on the stats/flags/NPC character insight accumulated in step 2).

The aim with this kind of game is a ~45 minute playthrough with high replayability. I also want a game where it is basically impossible to follow a walkthrough since the randomised storylet swarm is too unpredictable. The intent is for the player to enjoy exploring during the casual conversation in part 2, but know the higher tension stage 3 depends on it in unpredictable ways. I saw someone in a GDC lecture recently say that Sid Meier quote "Games are a series of interesting decisions" is better updated to "Games are a series of interesting consequences" and that really resonated with me. I enjoyed games like Civ a lot, but sometimes you don't want to spend the afternoon min-maxing percentages, you just want to intuitively feel your way around a complex game space and find out if the vase shatters when you push it over the edge.

I have drafted a smaller model storylet swarm conversation simulator in Twine, and will play-test it shortly to check how the writing holds together from the player perspective (writing dialogue that branches smoothly into the past and future is a fun head trip). Real-life, casual conversations strike a balance between following a narrow narrative flow, while always presenting the option of veering off to a new topic, and I would love to find a way to capture that feeling in a game.

Questions:
1. Has anything like this been attempted before?

2. As a relative newbie to renpy (but fairly experienced with writing big works of linear fiction) am I taking on a project that is going to eat me alive? (Drafting 80k words isn't stressful to me, renpy tutorials have been very straightforward, I'm planning on integrating simple public domain classical music myself and will insert placeholder storyboard graphics during construction, to be replaced by me or a to be determined artist collaborator when I reach the end phase of production- if the writing/mechanics are worthy of them). I hope to get a mostly finished game together in a year working on it half time.

All thoughts and feedback welcome. Swapping fiction critiques for the last few years has thickened my skin nicely.

[Fuseblower]

Interesting.

The combinatorial explosion is, of course, always limited by the number of story endings. This means it's pretty much a given that several routes lead to the same ending.

Keeping such routes a rewarding experience despite leading to the same result can be a bit challenging. Perhaps there could be a mystery that a player can only figure out by exploring different routes (even if they lead to the same ending). Or an ending could be "flavored" by the route traversed. For example : in a detective story, the apprehended criminal can have different motives depending on the route taken. And speaking of detective stories : the game "Cluedo" ("Clue" in America) has six suspects which would mean only six endings were it not that a murder weapon and a room is thrown into the mix which leads to 324 possible solutions.

I think this is what your "storylets" amount to. Combinations of storylets leading to other storylets and choices.

The visual novel would become quite procedural. Conditionals have to be attached to storylets. For example : a storylet X could have the condition that both storylet P and Q had to have been visited previously. Such things could become arbitrarily complex (for example : storylet X if ((P AND Q) AND NOT R) OR (V AND NOT K)) ).

Before you know it you'll be working with truth tables and Karnaugh maps to keep things logically consistent. That might be a bit much though

There's also the work of Vladimir Propp. His "Morphology of the Folktale" (1928) was meant to categorize all the different folktales but can also be used to make them. His research found that there were certain structural elements always present in these folktales and that they only differ in their implementation. There's a lot on the internet of his work. A structure like that can serve as the "scaffolding" of the visual novel.

Speaking of "scaffolding" : a very simple one is the structure of the typical 80s slasher movie. We got a couple of stock characters (the jock, the nerd, the prankster, the virgin, etc.). Their fates are basically set in stone and only the "final girl" (very seldomly "final boy") survives.

What if the choices the player makes will decide the character of the MC and with that, the role of the MC in the story? The "replayability" is then provided by exploring different roles in the story. There is the funny "Sorting Algorithm of Mortality" (tvtropes) where your chances of survival are decided by what kind of character you are.

So, instead of establishing main secondary characters (your point 2) the player could BE one of the main secondary characters.

Anyway, those are my two cents.

[Counter Arts]

One paradigm that players/readers enjoy?

The more they master/understand the story, the more that they can bend it to their own will.

Complete mastery of the story means anything theoretically possible within the story, they can make happen at will.

The whole excluding + randomized choices is something I experimented with before and players deeply hated it. It felt like their understanding of the story didn't help them navigate it iirc.

Imagine rewinding time only to find out a completely different set of events happening immediately once you entered the past. It doesn't make time-traveling as enjoyable as it otherwise would be.

[HaldaneBDoyle]

"This means it's pretty much a given that several routes lead to the same ending. "

I am hoping I can avoid this structure and stick to true branching in phase 3 (leading to 10-20 endings in my estimation). The story I have in mind is a tragedy (everyone dies unavoidably) but different paths change who dies when, and what good they manage to make of the tragedy for the wider world (and a little optional romance). Cluedo is a good analogy, but in my case it is just a matter of who dies when from a small cast of four characters (trapped in a doomed spaceship).

"Conditionals have to be attached to storylets."

This is the one thing I didn't want to happen (to steal a Brass Eye quote) otherwise the complexity of phase 2 will become unmanageable. The phase 2 conversation simulator is intended to accumulate conditionals for phase 3, but not apply them to the conversation itself. In my mind this works since the player character is pretty meek and agreeable. When they are having their first proper conversation with a new NPC they make a mental note of things they hear, and build up attitudes toward each other, but aren't really confident enough yet to say or do anything definitive in that moment. Only in phase 3, when high stakes events are unfolding, do those accumulated conditionals come into play.

It is also worth keeping in mind the conversation in phase 2 will almost entirely focus on the two character's pasts (as a way to flesh out the characterisation of both). To me this means that "plot" doesn't really exist in this phase since everything they are discussing is already resolved. To the characters there are no stakes beyond "does this person like me? (but the player knows information gathered here will influence the ending options). The only decision the player needs to make is which topic to chat about next. As a novel writer I often do this kind of exercise of "interviewing" characters to get inside their heads, understand where they are coming from before the story started. I thought it could be fun to make an interactive version of this that the VN player can steer.

I appreciate you bringing up the question:

"What if the choices the player makes will decide the character of the MC"

-but I am pretty sure I want to keep the player character as a fully authored personality (not a blank slate for the player to decide the personality of). Instead their capacity to act differently in phase 3 depends on how they have interacted with the other characters in phase 2. I also want to try to make the other characters have plenty of agency, which can often be catalysed by comments/reactions from the player character accumulated during phase 2. How much you trust or like another person can radically change what you expect from them under a high-pressure situation. I think the mechanic I want to play around with is almost like a butterfly effect of small seemingly insignificant moments having big consequences later. An option to make such connections explicit (lift the hood on the game engine logic) might be of interest to some players (but I'm not sure I want to write a story that is like a game of chess where the player is supposed to learn/master some arbitrary rigid rule set- it is a story to be experienced in varying ways). Real life isn't a game with simple, predictable rules. Maybe I am wrong and VN readers don't want that level of fuzziness.

[HaldaneBDoyle]

One paradigm that players/readers enjoy?

The more they master/understand the story, the more that they can bend it to their own will.

Complete mastery of the story means anything theoretically possible within the story, they can make happen at will.

The whole excluding + randomized choices is something I experimented with before and players deeply hated it. It felt like their understanding of the story didn't help them navigate it iirc.

Imagine rewinding time only to find out a completely different set of events happening immediately once you entered the past. It doesn't make time-traveling as enjoyable as it otherwise would be.

This is a really important point and I am very grateful you brought it up.

To me it gets to the heart of the difference between a game and a story. A game usually involves an attitude of rule legibility/understanding/mastery. A story usually involves a more passive attitude of acceptance/immersion/imagination. Interactive fiction tries to do a bit of both at the same time and that can cause contradictions.

I would maybe point to games like "What Remains of Edith Finch" and "Before Your Eyes" where there was no rule set to master, just a story to interact with as it unfolded as a way to achieve higher player/reader immersion. That is probably closer to the feeling I am aiming for with my project (though with more varied paths for replays).

I'm also strongly considering there being no "reverse your decision" option. In the conversation phase that would be a bit of a pain to code since there are multiple possible paths from any storylet both forward and backward. The short play time (aiming for 45 minutes max, so equivalent to an episode of a TV show) might be salient here as well. I would never implement any of these ideas in a VN that stretched for tens of hours of invested player time.

I would bring up another human activity that we seem to find extremely compelling- gambling. The roulette wheel has rules and mathematical probabilities, but in the end every spin is exciting because at the heart of it is something unpredictable. Though in a way I imagine that a player can achieve a degree of limited mastery of the game with replays, if the game is explicit about "key moments" in the conversation that influence the decision tree in step 3 of the game. But you cannot guarantee that those key storylets will come up in any particular playthrough since conversation is semi-randomised, so the player just has to make the best of the opportunities that fall into their lap on any particular playthrough. That isn't too different to randomised loot drops in an RPG which influences player stats and progression.

There might be a few "rare" endings that rely on a mix of skill (knowing which key conversation topics to steer towards), and luck (the RNG of the conversation engine going your way). That is the only way I could argue I am addressing your point about players wanting to be rewarded for taking the time to master the game.

All in all- excellent comment and it got me thinking more clearly about some key issues in designing the game/story. Thank you!

[Counter Arts]

Good to give you at least a heads up of what's ahead.

Though I should at least bring up old but interesting edge cases. Really old fiction that was interactive. And a sub-gerne of stories with zero-interactivity that almost played like a game.

Ancient plays actually was more interactive. If the crowd cheered hard enough of a character's death, the actors would give in and ad-lib the rest.

There used to be the golden age of detective fiction that was pretty close to a zero-interactivity game. The game was to solve the mystery before the detective does. Even had a kind of "unwritten ruleset" codified by Knox's "Ten Commandments" (or "Decalogue"). Know's Decalogue was kind of a deconstruction of the ruleset that many of those stories followed. Also probably why people really hated the culprit in "Heavy Rain" as it violated one of those rules.