The Mary Jane of Tomorrow

mj_of_tomorrow.jpgFor IF Comp 2015, I offered as a prize to contribute a piece set in the same universe as the author’s game. Steph Cherrywell chose this prize for Brain Guzzlers from Beyond!, which was exciting, since I’d enjoyed Brain Guzzlers a lot; and also slightly daunting, since Steph obviously didn’t need any help in coming up with art or feelies. Originally I was going to write a short story, but as I replayed the game and reviewed transcripts, I was hit with an idea for something more interactive. The result is The Mary Jane of Tomorrow, a not-too-difficult parser puzzle game set a few months after the events of Brain Guzzlers. (Estimated play time roughly 45 minutes, give or take.)

In the tradition of fanfic, it focuses on the relationship between a couple of the characters in the original game, Mary Jane Minsky and Jenny Yoshida. In canon, their closeness is demonstrated in various ways but never given center stage.

Gameplay-wise, The Mary Jane of Tomorrow is about training a robot to demonstrate certain personality and knowledge traits. To do that, the game makes extensive use of procedural text, borrowing the text generation library and even some of the corpora I used for Annals of the Parrigues. After the fold, I’ll talk about Mary Jane as a procedural text project, but it’s spoilery, so you probably want to play it first if you think you might enjoy it.

Steph decided she wanted to share her prize with the public, so The Mary Jane of Tomorrow is now available to play — and she even very kindly made some cover art for it, to match up with the rest of her work.

The game’s been uploaded to the IF Archive; in the short term, there’s also a Dropbox link for it, which I’m hoping will hold up until the file moves out of Archive Pending.

For some time I’ve been experimenting with a text generation system that makes use of a bit more persistent state than Tracery, but otherwise shares a lot of the same characteristics, especially the ability to nest symbols for complex expansion. When I considered a game about training a robot, this felt like a natural fit. The player’s actions would change the tags that the robot was using to select dialogue, and then it would dynamically produce output that matched its current training profile. Some of the outputs the game needed were:

Procedural recipes (requires cookery training): The first procedural thing I did in the game was create a generator for 1950s recipes, based off some websites that divulge the horrible secrets of cookbooks past: “Spam and Pimento Olive Surprise” or “Chipped Beef Loaf in Jello,” this sort of thing.

Procedural beat poetry (requires poetry training as well as the ennui attitude marker): In Brain Guzzlers from Beyond!, there’s an extensive puzzle around making poetry that will satisfy (or irritate) a beat poet. She gives a bunch of examples of what that kind of poetry will look like, which is itself pretty much a satire of the form. So I started experimenting with what a generator would do if it produced something similar to this.

Procedural romance couplets (requires poetry training and rhyme training): If your robot is trained in rhyme as well as general poetic ability, it will generate couplets based on cod Byron. I have a love-hate relationship with Byron: some of his poetry I quite like, some I consider nearly intolerable. To a Beautiful Quaker struck me as particularly susceptible to procgen satire, since there’s not really a strong through-line development of narrative or metaphor, and mostly the pairings just needed to rhyme.

Procedural limericks (requires rhyme training but a lack of poetry): these were tricky to do. Limericks require both rhyme and meter, and I turned to Twitter to suggest some word corpora with metrical information. Two sources proved especially useful: a corpus that lists words according to foot type (all the anapests you can eat!) and Allison Parish’s scripts for getting rhyme and meter matches out of the CMU pronunciation dictionary. I was also recommended this source of prosodically annotated corpora.

Ultimately, as interesting as it was to work with the constraints of metrics and rhyme, I feel like limericks were too constrained to make the most of procedural serendipity. It was possible to remix the first line a lot of ways — the first line of a limerick is usually basically a throw-away that just establishes the rhyme scheme for the rest of the verse. There was once an old man from Peru. But the rest of the limerick has to get a lot of work done in a short time: set up a situation, elaborate that situation in an inner couplet, then resolve it with a punch line. Getting good results required exerting so much fine control over output that it didn’t feel that procedural any more. I may take another run at this problem sometime, though.

Procedural freeform compositions (poetry training, no ennui): if the robot doesn’t know about rhyme, does know about poetry, but has low ennui, it will compose poems of three to five lines based on a couple available themes. One of the standard structures is Address-Simile-Proverb, so one gets output like

Listen up, creditor
Perform ecstatic dances at the altar of my wealth!
Shrouds have no pockets.


Beware, dearest soldier
Thy heart is a coin!
A sum entrusted to many servants diminishes for the master.

The proverbs are scraped from various online proverb lists, though I also had to cull the list a lot: when I first dropped the raw text in and fired up the game to test it, I was horrified to be presented with a poem whose last line featured a racist slur. Turns out people have come up with some extremely icky proverbs! I’d already done some reading about creator responsibility for procgen output, but this was a vivid reminder.

Or, in a slightly more complicated example:

Listen up, creditor
Pickle my wallet in vinegar!
The best things come in potting jars.
I will not complain.

In this case, the selection of “creditor” in the first line set the theme of the poem to money, which determined the choice of “wallet” as the thing to be pickled in line two. The fact that the system picked “my wallet” rather than “your wallet” then added “I will not complain” to a stack of possible concluding lines; and the selection of the pickling instruction (as opposed to some other instruction) added “The best things come in potting jars” to that same stack.

I don’t claim that these are great works of poetry, but I felt that, in comparison with the limericks, they more often came up with an interesting juxtaposition that I hadn’t already thought of myself.

Procedural similes: to feed both general speech and some of the poem forms, I have a subsystem that makes similes. Some of the simile corpus is pulled from web pages that list common idioms. The rest is based on searching BYU n-grams lists for adj-noun pairs such as “cold case” or “hot coffee” and then whittling that list down to the ones that struck me as the most memorable, proverbial representatives of their chosen adjectives.

Filters: the other thing I did was falls into the category of Cheap Tricks. The text converts “ing” to “in'” if the robot has learned about cowgirls; “the” to “ye” if the robot has read Ye Goodwyfe’s Bodice Overfloweth. There are drop-in alternatives for some other very common phrases.

21 thoughts on “The Mary Jane of Tomorrow

  1. Hi! Haven’t played the game yet (nor read the spoilers after the cut) — just wanted to say that I noticed that the Play Online link was broken (presumably because the server is unable to get past Dropbox’s login nonsense, which I imagine to be 100% intentional on Dropbox’s part), so I took the liberty of adding a temporary link on the IFDB page to a copy I’m hosting myself, which I plan to delete once it shows up on the Archive (or as soon as you like, of course). I hope that this is okay!

  2. Any other players figured what to do with the combination? I thought it must have been n pregnva oveguqnl, but I scrolled all the way back up and there wasn’t any further detail mentioned about that, so clearly I’ve gone chasing down the wrong garden path!

      • Oh, indeed, many times. I think there has been a misunderstanding — but it was an *enlightening* misunderstanding, I think, as it implies there’s a physical thing for me to find? In the parlance of lock and key, I have the former and some vague ideas about the latter

      • Yep, after having her read the “security” book (the choice of which greatly amused me!). Her suggestion is what sent me down the original path of trying to find out if the ovegugnl was described in more detail at the beginning, though I admit my *first* thought was to take her literally and see if the agricultural report mentioned anything :P

      • …ah, I figured it out. I wasn’t able to get the “key” because I’d pressed the button early in the game. Welp!

      • So you can! I could have *sworn* I’d already given her that, but apparently I hadn’t actually done so in my second playthrough. [facepalm]

      • There’s a cvar ahg ercbeg? Gosh, that would come in handy! But I’ve ybbxrq va naq haqre fb znal cvrprf bs fprarel gung gur cnefre vf abj tvivat zr fanexl zrffntrf nobhg vg (“Lbh’ir nyernql sbhaq zber guvatf ghpxrq haqre bgure guvatf guna lbh fubhyq rkcrpg va n zbfgyl rzcgl fubjebbz”), naq V unira’g sbhaq vg.

      • Fb V unira’g. Frrzrq n ovg vaqrpbebhf gb cerff gbb uneq ba gung sebag, ohg creuncf V fubhyq tvir vg n gel. Thanks for the nudge!

  3. So I’ve now played through this in full, and really enjoyed it. The story feels both charming and plausible as an extension of the original, and I really like the continuity nods (e.g., the zbfgyl boyvgrengrq fpvffbef) and emotional grace notes (e.g., the nqqvgvbany pbairefngvbany pubvpr jvgu Wraal gung nccrnef nsgre lbh’ir grfgrq gur npgvir yvfgravat srngher ba gur obg).

    The central puzzle feels very mathematical to me. In principle, of course, any IF puzzle is a mathematical puzzle, so this impression may just reflect what kinds of things I’m used to modelling mathematically, rather than any inherent property of the puzzle itself. (V jbex jvgu ovanel srngherf n ybg va yvathvfgvpf, naq zber fcrpvsvpnyyl, V erpragyl ernq n obbx znahfpevcg gung gerngf srngherf nf shapgvbaf gung nqq guvatf va be svygre gurz bhg (cbgragvnyyl inphbhfyl).)

    The procedurally generated text came up with some nice similes (I think there was something along the lines of “pbbyre guna n pnfg-veba cevil ba gur abegu fvqr bs n tynpvre”), and some of the longer poems were quite impressive. One thing I tend to find with procgen in general, not specific to this work, is that even when there’s lots of variation in the content, the repetition of templates and forms can create an overriding sense of sameness. In this work, of course, the story provides a good reason for the conversation to be a bit repetitious, and trying out variations on a theme is part of the fun. But I also wonder about what kinds of things might manage to get away from this property of procgen. I think further state-tracking might do it, especially if the generated text had more material consequences for the direction of the story, so that the choices made by the RNG could accumulate into more elaborate narrative structures.

    • So the reason I developed the procgen library in the first place is to do procedurally-modified character dialogue, where there’s still a lot of hand-authored text but the procgen is doing the job of giving emotional color appropriate to the situation. This is something we really needed for Versu, and that I feel I need to solve for the next system I do post-Versu. But to develop the library, I’ve been putting it through various paces in different types of projects.

      I did also think about a mechanism in this game where you had to get the bot to produce a poem with some additional thematic characteristics to solve a puzzle, or where you actually got it to write something that would let you communicate to Jenny, but ultimately it felt a bit out of scope and weird — storywise, I wanted the arc being about *you* learning to talk to Jenny better, not using your bot prop to do it all for you.

      But. Yeah. Other possible applications abound.

    • The proverbs are just random, not thematically tagged (except in that there are separate corpora of medieval proverbs vs. western sayings vs. advertising advice and so on).

      The words used in the first and second lines of poetry were tagged by hand. It might have been possible to do something more advanced, for instance by looking for words with shared associations via wordnet, but I felt like it would work fine to have a fairly small selection of hand-picked words in this context.

  4. Pingback: Visualizing Procgen Text | Emily Short's Interactive Storytelling

  5. Pingback: Mid-June Link Assortment | Emily Short's Interactive Storytelling

  6. Hey, it would’ve been good to know about those metrical corpora when I was writing the scoring code for the quasi-treasure hunt for the UK launch of Randall Munroe’s Thing Explainer. I set myself the task of detecting passages of iambic pentameter which might be embedded inside other text with the line divisions not marked; the saving grace was that the wordlist was limited. Anyway, I started from the CMU pronouncing dictionary and whipped my own up, which erred on the side of clemency when it had to err, which was a lot of the time.

    Here was the result: it’s not very good as showing my working out goes, but perhaps it’s useful or interesting anyhow.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s