Mad Lib Your Own Adventure

Continuing the discussion from Spam Watch:

Sorry for cross-linking from a Lounge topic, y’all.

As requested by OtherMichael, here is a cross-linked post, open to everyone. Putting this in craft, I guess. It could be a game… eventually. Meh. :confused:

Summoning @OtherMichael @daneel

So, a procedurally generated framework, perhaps turn-by-turn, perhaps generated from a seed (so always the same elaborated framework when generated from a particular seed) that the participants take turns to flesh out, perhaps mad-lib style, perhaps all in one go by the player who’s turn it is. (Lot’s of leeway in there for collectively constructing the scenario/particulars).

This topic is just for the purposes of discussion, throwing shit and seeing what sticks. :hankey:


Just played cards against humanity for the first time around the New Year, perhaps gaming the concepts to be used in constructing the story-fragment could be a sub-game in each round?

Oh and by:

I just mean, the most bare and basic type of indication. You llive, you die, you get the girl/boy, you find the treasure/macguffin. Not fleshed out at all; basic, archetypal, story particles.

I have done some rudimentary sketches of text generation ideas very much of a “mad libs” sort of idea. But I am not much of any sort of real coder, so the next steps of trying the ideas in code are much larger for me than the conceptual start.

What I was thinking of starts with a sort of database of words, where each word can have certain data associated with them. Such as what parts of speech / word classes they each belong to, weightings for the likelihood of position in reference to words of other parts of speech / word classes, real world categories, emotional weightings, rhyme, rhythm, syllables, and other such things. Then make a sort of tendency template which only yields loosely “plausible” patterns of speech forms, with consideration of word types and sentence lengths. And from there a process whereby I can specify some of these patterns, or statistical variations of them, along with maybe a stockpile of words to start with, and have it fill in the rest. Almost exactly like mad libs, except I need to give it a lexicon and some grammar first.

As a programming task, I don’t think it seems difficult to do well enough to get entertaining results. But a bit tedious. Realizing it is the sort of thing I need to psych myself up for and then just immerse in it.

1 Like

This (potential) project could be a great opportunity for us me to learn these types of coding. I’m game. heh

1 Like

Quick googling got me here.

I’ve seen a few games written with Twine, that might be the sort of thing you’re talking about, but I don’t know about the procedural generation stuff…

1 Like

Closest thing in that thread seems to be [Versu][1], which still wants to impute a great deal of detail into its working model.

My conception would be more of a history of previous actions, informing options presented to the players to flesh out, within the framework of archetypal story development.
I think this would mean some capability on behalf of the system, or perhaps the GM, to model the developments of the story tellers into the story-model so that relevant options are presented.

CYOA would essentially, if I remember correctly, present you with situational choices like go left or right and battles which you would conduct with die rolls and score accordingly.

My proposal would be closer to the bare bones of a CYOA novel, hiding the potential developments from the players (in the unread pages of the book, to keep with the analogy) whilst presenting some kind of gamified story telling option to the players.

The real problem seems to be how you would then insert the story developments in to the engine so that recognisable and sane story developments were presented.

Or perhaps a given result would be required, 'You attempt to bargain with your captor, you fail. How do you proceed?"

“I slay as many as I can before I am cut down and stand as an example of heroism to my people in the hope that they will take heed of my actions and rise up against this scourge.”

Result: Character dies, NPCs die, future characters of Players tribe are bolstered with +5 belief in themselves, 5% less chance of being ambushed. Reset character arc, begin new story thread.

or something.

The more I think about this the more difficult it seems.

Probably making the engine as absolutely simple as possible and having a GM pre-flesh-out options would make more sense but then, why bother with the engine at all? Gah.

Generating one story is the goal of Computational Narratology (herafter, CN) – and after 40+ years its still struggling.

Doing a CYA involves a lot of stories - each potential “decision point” branches off into a new story.

Now, real CN is complicated, but there’s no reason things can’t be “faked” and made more simply. That’s what a lot of the NaNoGenMo projects end up doing – finding “shortcuts” that may not pass rigorous academic muster, but produce relatively decent (or amusing) output.

A key insight into creating such works is not knowing the known limitations – because while the academic/scientific limitations may be well-defined, short-circuits are not-well-defined, and often turn out to be awesome. Try to do the impossible. Learn to code in order to do the thing everybody says can’t be done, and you might be able to do (or a simulacrum of same).


Here is my (still-in-progess, with a lot of gaps) templating (mad-libby) Fairy Tale generator

Here is some sample output (using the barebones preset); it gets quite long for two main reason: embedded stories, and “boss-fight mode” (which has NOTHING to do with the supposedly-underlying Proppian narratemes) which introduces a new villain when one has been killed:

Body is limited to 32000 characters; you entered 38274.

Well. Dang.

It’s over here

I must have heard about Twine in the distant past, but only started looking at it in November. I am interested to see how I could use it for experimental texts, and for generated output. There were a coupld of Twine-products for NaNoGenMo 214 – 2 of which were painful to read, and one that was somewhat interesting.

The author noted:

it was my first time doing this event and my second time really using
python for anything, and it was a lot of fun. this has me excited to do
more word generation stuff!

So have at it!

1 Like



Groovy and I guess the framework you have modelled in order to produce a functional story that adheres to archetypal fairy tales is what I’m getting at.

Gamifying the actual story telling through the participants, in a kind of… round the fire, turn by turn story and then imputing the results into the underlying model so that it presents relevant options.

Hmm, 'round the fire.
We could call it ‘Shaman’. The shaman knows the participants and would tell them stories based on the situation the tribe/village found itself in. Relevant models that she knows have an affect on the people involved because of the shared subconscious of the group.

We need not have this level of involvement with each player because archetypal stories have been hammered out and perfected over thousands of years. We already know what kinds of story are affective. We need only set the boundaries on potential development and allow the players to flesh out with interesting plot points.

So it would have to be turn-by-turn to allow for any direction changes the players introduce.

Who says it can’t be multi-genre…?

V interestiodad.

1 Like

Another track.

Let’s assume we have a fully working archetypal story engine which can take into account digressions, genre changes and unique developments.

How could we gamify the actual story development?

This is where I thought of Cards Against Humanity.

You have drawn a number of options and can use them to create the best result when inserted into the pre-ordained situation.

Mad Libs (which I have never played so I am assuming) is where you each take a turn to develop a sentence without knowledge of the prior developments?

What other crowd sourced story/situation development options could we use?

Now we’re paying Card Hunter or Hearthstone.

Mad Libs (and I suspect, Cards against Humanity) is about making dirty jokes and gets boring quickly.

Speaking of which, have you seen the Oatmeal’s card game on Kickstarter? Making millions…

1 Like

I was hoping we could pay the authors of Gilgamesh or The Bible instead. :smile:

Prior art for Shamanic method?



Yeah, that too. Stupid phone.

Kinda hoping someone (@funruly?) does a Badass dungeon crawler. A card-based mechanism might be interesting. But with a limited number of rounds, might not be the most workable method. And too much heavy lifting to be done off-site.


I wonder just how much functionality could be implemented into the BBS.

You could write a full CYOA novel, paragraph by paragraph, choice by choice into many comments of a topic, then reference those comments by quoting into the game thread.

Polls would allow you to select courses of action by the group… It’s silly but as a thought exercise, there must be a lot of different uses the BBS could be bent to.

This topic was automatically closed after 754 days. New replies are no longer allowed.