alt.hn

1/13/2025 at 12:35:22 AM

Standard patterns in choice-based games (2015)

 https://heterogenoustasks.wordpress.com/2015/01/26/standard-patterns-in-choice-based-games/

by Ariarule

1/13/2025 at 2:36:52 AM

This is great. I've been a game dev for about 30 years, much of which I've spent working with narrative design/writing teams. One thing I've learned to watch out for, especially among junior designers, is what the author labels the "Time Cave."

Narrative branching, done well, is fantastic—it gives the player agency and lets them make the story their own (as it were). But when you're creating the story graph, it's easy to get lost in it and lavish care on one path at the exclusion of the others. You can easily end up with one or two long, greatly-detailed paths, and (because dev time is finine, and you need to move on to writing other parts of the game) a pile of other paths that are shorter and less interesting. If the player takes one of the shorter ones, they end up missing out on all your coolest stuff. The tools I would design for the kinds of games I created specifically made it easy to create a main story trunk with side paths (that rejoined the trunk), and more difficult to branch/loop/etc.

Of course, that's not the only (or even the best) way to do narrative design—Disco Elysium is a masterwork because it did the branching, merching, loops, jumps, random checks, and so forth, so well!

by dejobaan

1/13/2025 at 4:46:04 AM

Your games rule :)

by spencerflem

1/13/2025 at 6:56:57 AM

What game is it?

by esperent

1/13/2025 at 8:32:21 AM

https://en.wikipedia.org/wiki/Dejobaan_Games

Played so much AaaaaAAaaaAAAaaAAAAaAAAAA!!! back in the day, still never 5 starred everything. Holds up IMO. Their other games are cool too

by spencerflem

1/13/2025 at 1:09:23 PM

You are too kind; thank you!

by dejobaan

1/13/2025 at 6:20:10 AM

I think it's a mistake to try get a story-focused game to have branching paths, akin to the old choose-your-own-adventure books. Until LLMs can proactively create new stories for the player to enjoy dynamically, i think it's always fraught with peril that the player fails to get the full story (or have to repeatedly play it and choose something else to try).

My personal preference would be to have a single, on rails story, where the player don't truly have a choice. It's an interactive movie.

Or, pick a sandbox mechanic, and let the player do what they want directly, and compute the consequence (the most common type being the physics system).

by chii

1/13/2025 at 8:49:44 AM

Humorously, this comment takes a giant swipe at 50 years of CYOA and Interactive Fiction.

There are over 14,000 games listed on https://ifdb.org.

Perhaps you should play some of them and adjust your perceptions.

by ChicagoDave

1/13/2025 at 3:30:03 PM

Also node the 52,000 visual novels at https://vndb.org/v .

by trothamel

1/13/2025 at 9:34:09 AM

I'm the opposite, apparently. I loved CYOA books as a kid because they could be reread, so I ended up seeking games that boasted multiple endings, including "bad" endings. When playing more linear games, I appreciated them for what they were, but there was a disappointment that I could not try different options along the way.

I think both have a clear place in gaming, since different gamers obviously look for different things.

by 0xEF

1/13/2025 at 8:50:20 AM

> My personal preference would be to have a single, on rails story, where the player don't truly have a choice. It's an interactive movie.

I can't bother to play those kind of games. A movie will be able to deliver its stories a lot better than a game.

But with choice and branching you get to appropriate the protagonist(s) and some story events can be a lot more impactful then. Lately I played Cyberpunk for which you have some choices in most missions and the endings hit different. If anyone involved in the DLC story is around: kudos to everyone involved in making the "face in the crowd" ending. You play some almost super heroic character and due to your choices (which involve betraying and killing a lot of people) you get to survive: alone and back to generic human power level.

by arkh

1/13/2025 at 9:19:25 AM

> A movie will be able to deliver its stories a lot better than a game.

I don't agree. Something like SOMA would just be a generic sci-fi B-movie but it's an awesome game, even though there's no real choice and is in essence just a walking simulator.

by Sander_Marechal

1/13/2025 at 7:48:19 AM

If you're not going to have choices matter, why make the story in an interactive medium at all? Branching paths require a lot of compromises, but there are still things you can do much better with handwritten stories than in a sandbox style.

by lmm

1/13/2025 at 8:40:12 AM

> why make the story in an interactive medium at all?

have you not seen the success of the COD Modern Warfare franchise? Their single player game is essentially an on-rails shooter, with pivotal story points completely scripted (you "press the buttons"). There's no choice, there's no branching (of the story).

But people like to shoot, like to run around, etc. It feels like they have control, and it feels like the heroics in the story is their contribution.

by chii

1/13/2025 at 8:47:17 AM

Halo is the same. It is essentially a very long hallway with enemies to take care of before you can move to the next hallway. Also, I recently played through the first Halo again and it was still quite fun.

by mnky9800n

1/13/2025 at 8:45:47 AM

I thought we were talking about a story-focused game, which that is not.

by lmm

1/13/2025 at 2:26:39 PM

You skip through the "pivotal story points" and ignore them.

by watwut

1/13/2025 at 9:10:34 AM

Isn't that dismissing 90% of games? The story can exist purely to give emotional context to the action of the game.

by zelos

1/13/2025 at 9:13:06 AM

I meant in the context of a story-focused game, which is indeed less than 10% of games in general.

by lmm

1/13/2025 at 6:49:37 AM

imo its not necessary to get the "whole" story, or replay it to see every possibility

just seing "my" story is enough

interactive movie type games are great but they're a different experience from choice style games which are also great, (and ofc the article shows that there's many styles within this too, all with a different experience)

I don't believe LLMs can recreate the same authored experience that has a point of view. I think they'll be okay at genre work soon enough though, for better or worse. But thats not a type of game I'm personally interested in.

by spencerflem

1/13/2025 at 4:52:32 PM

There a lot of great games with branching narratives, for example The Outer Wilds, Baldur's Gate 3, The Witcher 3, all the FromSoft games have extremely powerful narratives with multiple branches and multiple possible endings based on player choice. Done well there's a lot of replayability that arises from the desire to experience different paths within the overall general experience, as well as the immersion that comes from the player truly feeling they have agency in the world in the way that an on-rails experience just cannot do. For example there was one quest in the Witcher 3[1] where I remember being haunted by the consequences of an extremely difficult choice I had to make and wondering for days whether or not I had done the right thing. I eventually replayed the game and when I got to that point .... decided to make the same choice. sigh. So I still don't know whether I did the right thing. I just couldn't bring myself to try the other branch - it just wasn't right for me. One of the most powerful things a game has ever done to me, and very true to the complex and ambivalent moral tone of the books in my opinion.

Done poorly, you end up with something like Fallout 4. Theoretically there are choices but they don't matter that much and many of the branches get little to no QA love so have bugs and problems (eg when I did my playthrough of fallout 4 I unwittingly managed to avoid the main choice you're supposed to be forced into doing which is to choose which faction to align yourself with) which had numerous buggy consequences because the devs clearly had never expected anyone to play in this particular way (I basically did a breadth-first traversal of all the different quests but put off a couple of key decisions because I couldn't make up my mind. In the end this pushed me past the point where I was supposed to make the choice and I ended up friends with just about everyone. Except the institute. Screw those guys). Yet another reason that game was so disappointing.

[1] The whispering hillock. If you know, you know.

by seanhunter

1/13/2025 at 5:48:28 PM

While I enjoyed playing Fallout 4, I was really annoyed with how it forced you into a specific path that lead inexorably towards a very specific ending no matter, as you say, what you do.

At the end, you have just two options: You can join the institute, or make war with them. I actually really didn't care about this faction and wanted to get out of that storyline, but that's what the game offers you.

Of course many games lead you to a specific denouement, which is typically one where you've defeated all the bad guys and win. But those games don't make any pretenses about making choices, whereas Fallout does.

I was really impressed with how many choices you had in Baldur's Gate 3. They definitely recorded an insane amount of dialogue just to cover all the possibilities. But I did find myself annoyed with the ending here, too, which only allows two choices at the end.

If there are two choices to make, it's basically the same as one, because the story just stops there. I'd rather games stop pretending. Don't pretend that there are so many gray areas. Just let the good guys win over the bad guys.

An alternative approach for such a game would be to have the decision tree be oriented around a kind of moral compass throughout, and your deeds decide what you can do. So if you keep killing or betraying people, you become more evil, maybe physically deteriorating into a kind of ghoul, while normal people start to fear you and refuse to barter or make allegiances. But as you get more evil, you gain access to eldritch abilities, make friends with monsters, and so on. There may be a point where you can make amends and return from evil, or vice versa, but at some point there's a point of no return where the final trajectory has been decided.

There are some games that have some something like this, where you have a "reputation" among the good guys that you can lose by doing misdeeds. Not sure if exactly what I describe has been done, however.

by atombender

1/13/2025 at 6:07:06 PM

That alternative approach is basically what happens in Dishonoured and Dishonoured II where there is a “chaos” meter which determines how much the world descends into disorder based on your choices to kill people or let them live. There are multiple “clean hands” options and there are always ways to complete any particular mission or level with clean hands. There are some elegant ways the world’s chaos state are revealed to the player. For example in Dishonoured, the child Emily’s drawings[1] are more or less happy reflecting the chaos of the world.

[1] https://dishonored.fandom.com/wiki/Emily_Kaldwin#Emily's_Dra...

by seanhunter

1/13/2025 at 4:58:43 PM

> Done poorly

yep, and i argue that most game devs wouldn't do it well enough to justify it. The limited time and resources available means they're almost certainly better off not branching, but make one good main story and polish it.

Take a look at the examples you listed - they're all award winning games, from developers with serious experience, grit and determination to make the best game. They spent ages, and lots of resources to do it.

Even big studios, with similar or more resources, fail at making such branching good. It's a folly to think that a smaller game developer with more limited resources could make it better. Disco Elysium is almost an exception that proves the rule (or another example is Pathologic).

by chii

1/13/2025 at 4:58:07 PM

I felt that The Outer Worlds did quite a good job, too. From the devs who built Fallout New Vegas.

Funny little game. I find it quite charming in its eccentricities.

by Alex-Programs

1/13/2025 at 2:28:37 PM

I think a good idea is to have a tree-like checkpoint-save system so you can always go back to the state you were in before descending down into a branch and go down another one without replaying everything up to this point. It encourages replays and exploring all content.

Papers please had something like this.

by scotty79

1/13/2025 at 2:42:01 PM

and if i recall correctly, also Detroit: Become Human

It also tells you the % of people that managed to reach a certain ending or branch.

by chii

1/13/2025 at 2:25:00 PM

I strongly disagree. I was actively preferring those games and found it fun to try out different endings.

> My personal preference would be to have a single, on rails story, where the player don't truly have a choice. It's an interactive movie.

I stopped to play those games. Movies are better at being movies then games.

by watwut

1/13/2025 at 4:14:50 PM

> I stopped to play those games.

English nit, if you mean you no longer play those kinds of games, as I think you do baaed on context, you should write

"I stopped playing those games."

by toast0

1/13/2025 at 7:00:43 AM

I completely disagree, there are plenty of branching games which are extremely good and which would be severely worse if they weren't branching: Disco Elysium and Baldur's Gate 3 come to mind.

We are very very far off from an AI being able to come up with compelling stories that are logically coherent.

by suddenlybananas

1/13/2025 at 8:42:33 AM

> there are plenty of branching games which are extremely good

i think you mean there's barely any good ones. The examples you come up with are the exceptions that prove the rule. Look at a game like Dishonered, where the story have _some_ branches, but it's half-assed imho. There's plenty more games where having gone for a branching story made the game more expensive, less deep, and harder to sell as a result.

by chii

1/13/2025 at 4:45:41 AM

This is fun, and holds true in the creative writing group I run. We use a website I programmed that helps us collaborate on writing branching fiction. We have a mapping utility that creates graphs like in the article, except more animated (d3.js, elkjs).

As different authors can start their own new stories, one thing I often have to deal with is that they want to design their story to have both long path lengths (multiple chapters before an ending), and also high choice count. Those of you who know something about geometric series know that this causes problems. I often have to tell them they can't have everything they want, which causes minor drama. :)

As a result, one of our stories basically shot its "choice budget" in the first few chapters, leading to many linear paths in the latter parts of the narratives, which is fun in its own way.

Another of our stories has just started playing with the "gauntlet pattern" as the article describes. For this one, we decided that all chapters must be in the "same universe", just following different characters' perspectives, and are planning for certain "anchor chapters" where all characters come together for a meeting. Probably the detective questioning them as a group (it's a murder mystery).

All of our stories are supposed to be literary, so usually in third person, sometimes first, never the second-person. So we don't tend to use choices and chapters as directions and rooms; it's all about how the plot moves. We also don't track state; they're designed to be able to be printed as books people can page through.

Overall a super-fun project for me and a handful of other writers, it's been a consistent way to spend a few hours of fun each week.

by tunesmith

1/13/2025 at 7:42:22 AM

You can still track small state via the reader. I vaguely remember a choose your own adventure as a kid:

If you picked up the key earlier, turn to page XX

Otherwise, turn to page YY

It was entertaining to a) suddenly realize I had missed an important detail or b) allow me to “escape” if I can’t find the key or just don’t like that part of the book.

by withinboredom

1/13/2025 at 9:36:47 PM

I remember some books that were explicit about tracking the state. There was an old space adventure story I read as a kid where you wrote your stats out on a piece of paper:

    shields: 100
    lasers: 100
    troops: 100
    hyperdrive: 50
    days: 0
As you read the book it had you encounter hostile threats and even roll dice and do lookups on tables at the end of the book to decide what happens next. There would be places like "To hyperjump over the rift you need at least 40 hyperdrive points left, add 3 days and turn to page 88. To go around add 7 days and turn to page 41." There were even fights against enemy ships where you had to roll for both sides and do chart lookups to see how many troops you lost in the boarding actions and how badly the ships were damaged. You could even lose a fight and end the run right there.

The plotline involved you racing to some planet to deliver news of an impending invasion or something and if you didn't get there within I think 30 or so days the news would be too late and you would lose.

However, being the nerdy kid that I was I mapped out every single possible route in the book and then simulated all events going perfectly on each route and there didn't appear to be a single way to actually win. The author had not done the math right and the absolute fastest you could finish was like 40 days even if you got crazy lucky with the dice.

by jandrese

1/13/2025 at 8:37:27 AM

One book I read did a similar thing, but managed to do it spoiler-free. There was a magical crystal, I think, that did some magical things. When you pick it up, the book says "To use the crystal, look at paragraph (current paragraph + 20)" and the author actually managed to do that for the most paragraphs from then onwards

by dasfsi

1/13/2025 at 9:35:16 AM

An old gamebook series that did lots of this is Steve Jackson's Sorcery! -- I wonder if that could be what you're thinking of?

More recently, Jason Shiga has used clever mechanics like this a lot in comic book form, notably in Meanwhile. He's just finished a three-part series aimed at younger readers, Adventuregame Comics. All Shiga's stuff is great, highly recommended.

by iainmerrick

1/13/2025 at 8:39:36 PM

"Scorpion Swamp" has a map-style story layout, you are also encouraged to re-create the map on paper as you explore. This also means that you can get back to earlier locations where enemies will either respawn or rest on the second read-through.

by rzzzt

1/13/2025 at 3:36:54 PM

I usually read text IF following all paths at once, with heuristic combinations of breadth-first and depth-first search in order to maintain the unexplored front small. Not much different from drawing a map in a computer RPG.

by HelloNurse

1/13/2025 at 3:44:21 AM

Nier Automata is my favorite example of the relatively rare "Loop and Grow" pattern. You play through the game three times, with each iteration enriching and elaborating on the story and characters. Brilliant and weird narrative structure.

by quotemstr

1/13/2025 at 4:31:53 PM

Isn't Loop and Grow roughly the same as Metroidvania? Or does Metroidvania not require an actual repeating loop, whereas Loop and Grow does?

by twic

1/13/2025 at 10:21:54 PM

I picked up some TutorText books recently, a programmed instructional series from the late 50s to 60s. At the end of every section there's a multiple choice question, one answer leads to the next section, the other to an explanation of why that answer was wrong. Pretty pure Gauntlet, I don't think there's any follow up questions on a dead end path, though I haven't mapped them all out fully. I like the idea of tailoring explanations to specifically anticipated misconceptions.

Indexed here: https://gamebooks.org/Series/457/Show

And some exposition from Hackaday a few years ago: https://hackaday.com/2020/08/28/a-tale-of-tutor-texts/

It looks like I need to take a closer look, that last article says

> often the wrong answer pages take you on a detour path to correct your thinking before rejoining the main line of the book.

which is what I was hoping to find.

by hcs

1/13/2025 at 12:20:27 PM

I don't quite understand what "floating modules" are. Is it something akin to sidequests in a CRPG or a "sandbox VN"?

One interesting (and very complex) approach I've seen in VNs is multiple interleaved paths. Each path looks like a branch and bottleneck, but at certain points a decision taken on one path blocks or forces an outcome on another. You can linearize it into a single "branch and bottleneck" with extensive state tracking (this is how it's implemented, after all), but it's far easier to model it as multiple paths.

by orthoxerox

1/13/2025 at 3:10:26 AM

Reminds me of Disco Elysium Explorer[1]. Conversations 7, 8, 9, and 10 are great real life examples.

1. http://134.0.119.41

by kelseyfrog

1/13/2025 at 2:15:38 AM

This is very interesting, thanks for posting! Makes me think of the big choice diagrams in Detroit: Become Human. I wonder if there is any literature about this?

by flpm

1/13/2025 at 2:53:31 AM

I'm curious about Alpha Protocol. Probably the #1 game I've played for choices matter.

Detroit is interesting, in that it includes some choices made by passing or failing QTEs. They really did the "You will get emotionally stomped if you screw this up!" well in that game. I don't know its structure well, as I only played it once. (So experience only one path.)

Unless you count time caves like The Stanley Parable!

by the__alchemist

1/13/2025 at 3:13:54 AM

> I wonder if there is any literature about this?

Came here looking for the same.. some kind of map from the game design angle more towards game theory.

Fun semi related tangent, I was curious to know authors background, and the About page quotes Borges “garden of forking paths” which jives nicely with tfa. Cataloging rather than inventing is an underrated activity in math sometimes, and we need to do both. Game garden taxonomy!

by photonthug

1/13/2025 at 1:21:01 PM

I loved the Lone Wolf and associated books as a kid. One of my favorite things about them was seeing glimpses of possible story paths as I flipped around the book. When I would get to the end knowing I had missed some enticing possibility in the story line, I would immediately flip back to the beginning and start over.

by numbsafari

1/13/2025 at 5:07:49 AM

gauntlet looks like GTA V's story mode pattern.

GTA2 with it's competing gangs seem to have a "state" tracker in the form of reputation scores with the game, while having an open world map. As your reputation/state changed opportunities would become available/unavailable.

I still think GTA 2's system is impressive.

by Over2Chars

1/13/2025 at 8:29:14 PM

IIRC in one of the "Fighting Fantasy" books the hero is captured and can choose if they want to spend the rest of their lives in a standing-only or a sitting-only cage - the story ends badly either way. I was a bit angry reading it as I didn't have a "save state" earmarked with the index finger.

by rzzzt

1/13/2025 at 10:27:00 PM

> To avoid obliterating the effect of past choices, branch-and-bottleneck structures almost always rely on heavy use of state-tracking (if a game doesn’t do this, chances are you are dealing with a gauntlet).

I do not see that branch-and-bottleneck relies on use of state-tracking any more than gauntlet.

by chrisjj

1/13/2025 at 3:41:00 PM

This was actually a really cool analysis. Fun to see how these things can be broken down so cleanly into graphs.

by eugenekolo

1/13/2025 at 2:52:10 AM

This is fantastic. Does anyone have any book references that help you do writing in some of these formats.

by codazoda

1/13/2025 at 8:28:32 AM

There is this collection of 1980's internal design documents from Flying Buffalo that you can buy:

https://www.drivethrurpg.com/en/product/296847/t-t-solo-desi...

It is old, but I do not know of any other books really on this topic. I enjoyed reading it anyway. One of the documents in it describes a simple manual algorithm for how to number the sections in a reasonable way (just randomly assigning numbers is not very reasonable, as anyone will learn from reading a book where the author did that). I implemented that in a simple pandoc filter:

https://github.com/lifelike/pangamebook

by livrem

1/13/2025 at 2:55:36 AM

not sure if this perfectly addresses your question, but "Designing Games" by Tynan Sylvester is a great resource

by egglemonsoup

1/13/2025 at 6:33:55 AM

Nicely timed I’m making a CYOA game right now so very interested learn more.

by andrewstuart

1/13/2025 at 9:35:23 AM

Hi, I've just ‘finished’ an app to make interactive text stories/games and I'm looking for people to test my app with real stories, if you're interested I could help you with your story so you can publish it online wherever you want.

by ardleon

1/13/2025 at 5:10:38 PM

I’d give it a look my email is in my profile

by andrewstuart

1/13/2025 at 5:21:43 PM

ok, the only thing is that the application is in Spanish, at the time I did not contemplate that it was multilingual, among other things because I made it for me, if you are still interested I will write you and send you a link.

by ardleon

1/13/2025 at 7:38:21 PM

could chess be considered a choice-based game?

by yapyap

1/13/2025 at 6:02:07 AM

Ai can produce a new type of game where choices are dynamic and outcomes are generated by LLM agents. Fiction is an hallucination and LLMs are master hallucinators.

Basically LLMs have to be given assets and game components that they can easily compose.

by ninetyninenine

1/13/2025 at 10:36:41 AM

> Fiction is an hallucination and LLMs are master hallucinators.

They're jacks of all trades, master of none.

This has its uses, but they have limits, and for now at least, those limits are under the threshold for that.

I have actually tried using them to make a text adventure to help learn German. The result was at the lower end of the quality range I've witnessed from LLM output: a nice first draft, not shippable, missing a core element, missing a lot of content, too simple, the kind of thing where you'd give the output of the LLM as a code challenge to a job candidate to see how they improve it.

by ben_w

1/13/2025 at 9:56:47 PM

An AI can easily produce a filler slop story. They struggle much harder with creating something new and interesting. For a CYOA type story for kids that might be reasonable, although they tend to make a hash of the details. There are more problems, like does the AI know when to stop? Can it recognize or generate a bad end or a good end without explicit instruction from the player?

Something like:

Generate a choose your own adventure story about a young boy shipwrecked on an island populated by hostile pirates. A hidden cave holds treasure. There is a jungle on the island. Dangerous jungle creatures inhabit the island and the boy can not fight. Also hidden on the island is a boat. Each story section is around 200 words long and ends with a multiple choice question for the player to select which path they want to pursue next. The story is complete either when the boy dies or finds the boat, after no more than 20 story segments.

I have some doubts the AI will be able to handle all of that and keep it interesting and coherent. This sort of storytelling requires some attention to detail that LLMs usually struggle with.

by jandrese

1/13/2025 at 7:50:27 AM

Having had an LLM tell me a story, my answer would be that this is a dumb idea. LLMs have no concept of realistic cause and effect.

by withinboredom

1/13/2025 at 8:08:32 AM

Nah I think not considering this idea at all is the extremely dumb and brain dead opinion. LLMs can tell stories. Realistic causes and effects aren’t even consistent in human stories. A good story isn’t 100 percent dependent on this.

The LLM walks the line between hallucinating too much and sometimes not. Either way you can pretty much guarantee that almost all stories made in games now are already mostly written by an LLM. It’s just the writing is edited and curated by a human.

by ninetyninenine

1/13/2025 at 10:17:50 AM

I've done a short LLM-powered VN, and LLM actions were restricted to local interactions only because of how weak it is at making up the story. It's great at removing the parser-based interactions, but I think that's it.

There's a second technical problem that such stories are represented by a form of state-machine and that you would need to recompile it on the fly, making many checks very difficult (you would need to be able to check reachability on the fly, chunk transitions, etc). I think it would take years to get to the level of some of the great IF games with an LLM, and not just a cool PoC.

by woolion

1/13/2025 at 10:28:07 AM

As a grumpy old symbolic ai hand I do wonder if it was possible to build a (perhaps crude) ontology based simulation with consistency, cause and effect and so on and then use the results of that for prompting an LLM.

But as a consumer, I lean far to the side of "give me a handcrafted tunnel experience with the illusion of choice" in the divide between consequences yes or no. I don't think I'd actually want this "simulation behind an LLM facade". If I'm in the mood for reading (or for listening to voice actors reading to me), I'd rather have it be something more meaningful than just a game state. But to those on the other end of the spectrum, this might actually be the holy grail of game building.

by usrusr