The Stories We Tell

(and the tools and data that we use for them -

an investigation into "Plotto")

Lynn Cherny

@arnicas

(ex)-emlyon business school

What's a story?

"A sequence of events someone reports."

-me

Implications

  • Time passes and related events occur. (May be fictional.)
  • Judgment about specialness / importance occurs by someone - not all sequences are stories.
  • There is a creative description of "events" - this is the form (e.g., it may not be linear).

Once upon a time, there was _____. Every day, _____. One day, _____. Because of that, _____. Because of that, _____. Until finally_____.

Jon Schwabish: What Is Story? (5 Part Series on Data Stories)

Me: "Data Characters in Search of an Author"

Various posts, talks, and papers by Robert Kosara

(plus lots more)

A Little Good Reading on Data Storytelling...

Plan

  • Tools for story generation
  • A particularly problematic one & data analysis
    • WARNING: Some offensive text.
  • Wider thoughts about culture, data, story

Tools for Storytelling

Very broadly,

 

With a focus on "Fiction" -- books, movies, game story.

(data comes soon though)

A naive typology of tool types

  • Help you create the form (the output)
    • Comic-strip generators
    • PowerPoint and Word templates
    • Unity game engine, Adobe Illustrator...
    • Screenplay tools
    • etc
  • Help you with the content-creation itself (my interest here)
    • Creativity prompts
    • Clip art libraries
    • Inventory systems
    • AI / proc-gen data-trained tools

Supporting the writing

process might

require non-linear,

or visual tools,

even if the reader

gets a linear piece of 

text.

 

E.g., Scrivener.

"interactive fiction"

Queneau,

1967

Choose Your Own Adventure

Books (CYOA)

A text hypertext.

Space & Beyond

CYOA book

narrative choices

"macro vis"

A Tiny Sample of Inventory Studies of Story Elements

Booker's 7 Plot Types

  • Overcoming the Monster
  • Rebirth 
  • Rags to Riches 
  • Voyage and Return
  • Comedy
  • Tragedy 
  • The Quest

Propp's Motifs (from Fairy Tales)

VIIIa. ONE MEMBER OF A FAMILY EITHER LACKS SOMETHING OR DESIRES TO HAVE SOMETHING. (Definition: lack. Designation: a.)

 

IX. MISFORTUNE OR LACK IS MADE KNOWN; THE HERO IS APPROACHED WITH A REQUEST OR COMMAND; HE IS ALLOWED TO GO OR HE IS DISPATCHED. (Definition: mediation, the connective incident. Designation: B.)

Thompson Motif Index

Suppose we want to support story creation... using these inventory efforts?

Example use of motifs in story generating tool research...

An Incomplete List of People and Conferences related to digital storytelling, if you're into this!

"Plotto"

(Immense thanks to 

Gary Kacmarcik

for the hypertext

version I used in

this project.)

Special Transformation Rules....

Yeah, so I laboriously converted the HTML, made the network, wrote the regexs and string subs,  etc....

Lots of it is actually non-"standardized" and hard to handle and turn into "real text":

Thennnnn.....  I got this:

female protagonist, of an inferior race, rescues male protagonist, of a superior race, and falls in love with him.

"I'm a data person - why didn't I look at this data more closely first?"

🤔

So, there are 3 top level categories of conflict- "Group" in the table here:

There are subgroups.... but not if you're married.

Reminder of CYOA books - can be seen as networks:

Notation for a network:

Node:

Edges going to other nodes

Edges coming from other nodes (some with instructions)

(Degree 6)

Interconnections by Major Conflict Type

Married Life

Love & Courtship

Enterprise

Even with Married Life being a small conflict set, it's among the highest degree nodes...

Highest degree node?

B-0, rescued from an accident by A-0, whom she does net [sic] know, falls in love with him.

Story Conflicts by Binary Gender of Main Characters

(I didn't count the aunts, fathers, etc - they were minimally mentioned)

The "Bechdel Test"

Far fewer stories feature more than 1 woman  - many stories contain many men

1 story has 3 female chars, a few have 2, 

most have 1.

1 story has 4 male chars, a few have 3,

lots have 2.

the story with 4 men in it, icyc

"A-0 is a United States consul * A-0 lives on an island, and on the same island are two other white men, A-2 and A-5, both friends of A-0's * A-6, an officer of the law, calls on A-0 to help him arrest A-5 **"

Stories with only women.... pretty depressing and clichéd.

(A surprising amount of "impersonating a boy" and suicide. Well, maybe not the latter in this guy's universe.)

"B-0's friend, B-2, an attractive married woman, seeks to save A-0, B-0's fiance, from the wiles of a designing woman, B-3, and restore him to B-0. B-2 does this by winning A-0 away from B-3"

 

Stories with multiple women are usually all about the men. (The B's are female characters, A's are male.)

'A-0 is engaged to marry B-0. B-3, a designing woman, seeks to compromise A-0 (218a, b) so B-0 will give him up. B-2 is a generous woman who seeks by secret enterprise (844b) to rescue A-0 from the wiles of B-3 and restore him to B-0'

Multiple-

female

stories

seeks,

mother,

father,

order,

happiness,

life...

Multiple-

male

stories

friend,

seeks,

enterprise,

finds,

(other male

chars),

money

 

 

There are about 30 reprehensible "race"-related conflicts, e.g.:

B-0, of an inferior race, falls in love with A-0, of a superior race
A-0, of an inferior race, falls in love with B-0, of a superior race
B-0, of an inferior race, in seeking to win the love of A-0, of a superior lace [sic], learns how hopeless is the task of challenging racial conventions
A-0, in love with B-0, of an inferior race, seeks to abandon B-0 secretly in order to uphold a lofty conception of duty
A-0, with a taint of negro blood in his veins—known only to himself—loves and is beloved by B-0, a white girl
A-0, a white man cast away among bloodthirsty savages, has his life spared because he is a ventriloquist and supposed to be a god

 

A-0 finds himself the only white man in a tribe of half-savage natives * A-0, finding himself the only white man in a tribe of half-savage natives, is compelled to struggle against their primitive superstitions **

 

A-0, a white man battling against the superstitious frenzy of a half-savage tribe stricken with the plague, upholds the highest ideals of the white man’s civilization

 

A-0, a white man of brilliant intellectual attainments, battles for existence in an isolated, primitive, savage wilderness 

The "savages" aren't noble, they're stupid and dangerous.

Basically, I've had enough of Plotto.

P.S. Tools Used

  • Python & Jupyter Notebooks
    • Altair (charts)
    • NetworkX
    • Pandas
    • BeautifulSoup
  • Gephi
  • Neo4j (not shown here)
  • Sublime Text
  • Jason Davies Word Cloud Generator

What Cultural Stories Are We Telling Now?

Surely we know better.

As well as the drop in the number of characters who are women or girls, they also found “a fairly stunning decline” in the number of books written by women in the first half of the 20th century, writing that “the proportion of fiction actually written by women … drops by half (from roughly 50% of titles to roughly 25%) as we move from 1850 to 1950.”

macro analysis over time!

Women as characters, and gendered descriptions.

One of the main things men do have is a pocket; in the twentieth century they are constantly putting things in it.

The Pudding link -

by Jan Diehm and

Amber Thomas

On average, the pockets in women’s jeans are 48% shorter and 6.5% narrower than men’s pockets.

Me: This is why we need diverse creators of data vis and data analysis.  They will make different stories than you're used to.

Many of readers are drawing conclusions that were anecdotally obvious to women in the film industry. But nobody wanted to do the grunt work of gathering the data.

Crowd-sourced

tv & movie

"tropes"

visualized by

Bocoup

(Yannick Assogba,

Irene Ros,

Jim Vallandingham)

http://stereotropes.bocoup.com/

To be clear, the show about boys got way too much credit, and the show about girls got way too little. This is how we approach male vs. female work. Let’s call it the “male glance,” the narrative corollary to the male gaze. We all have it, and it’s ruining our ability to see good art. The effects are poisonous and cumulative, and have resulted in an absolutely massive talent drain. We’ve been hemorrhaging great work for decades, partly because we were so bad at seeing it.

About shows on HBO that got press & award coverage.

This is 

bigger than #MeToo.

Linda Bloodworth Thomason, one of CBS' biggest hitmakers, reveals the disgraced mogul kept her shows off the air for seven years: "People asked me for years, 'What happened to you?' Les Moonves happened to me."

Data and indignation...

you need the data, though.

Aaron Williams' talk at OpenVis 2018

"How Data, and the Visualization of It, Helps Us Understand 'Us'"

You can't study something

unless there is data for it.

E.g., "race" in the census.

But: Similar stories are told differently.

Washington Post: Unequal Justice

Racial differences in homicide arrests in USA

“When you look at white murders, those crimes get solved even when there’s not a witness. Black murders, it seems like we’ve got to have a witness to the witness before we get an arrest,” said Franklin.

(whose story is being listened to? what "data" is being asked for?)

What data gets used to tell stories?

Some guy's hand-made biased inventory system....

 

 

AI & "Big Data"-driven design?

OR, instead of some dude, what if we used

In the imSitu training set, 33% of cooking images have man in the agent role while the rest have woman. After training a Conditional Random Field (CRF), bias is amplified: man fills 16% of agent roles in cooking images.

Zhao et al. ("Men Also Like Shopping")

Generating Stories from Images?

Embeddings (e.g., word2vec / GloVe)

so super popular for text generation and classification!

“father : doctor :: mother : ?

One might have hoped that the Google News embedding would exhibit little gender bias because many of its authors are professional journalists

NURSE.

Char RNN's to generate text: Janelle Shane's work

My Students Tried With a Dataset of Vegetarian and Vegan Restaurants...

Tofe Mexican

Thin’s Mall

Natural Coffeeline Kitchen

Pango Garden

The Restaurant Grill

Datine Mexican Cafe

Pizza Pro

Pizzu Mexican Cuisine

Sumon Chicago

Taco Peate Inn

Pizza Postaurant

 

The Pizza & Cafe

Cafe Food

Food’s India

Vegetarine Pizza

Salad Cafe

Salad Star Cafe

Suba Pizza & Pizza

Chick Chick Chick Chick Chick Chicken Cafe

Wat?!

>grep Chick data/vegetarian_vegan_names.txt 

Lo-lo's Chicken & Waffles

Natural Chicken Grill

The Chubby Chickpea - Closed

Healthy Chicken Cafe

Chubsy Wubsy Pizza & Chicken

New Jumbo Fried Chicken and Pizza

Chick-o-pea's

Vitamin Chick

Natural Chicken Grill

Chick-fil-a Cedar Hill

Natural Chicken Grill

Chickpeas Vegetarian

Chicken Al Mattone - Frisco

Lebanese Chicken - Hamtramck

Chickpeas - Closed

Fiesty Chicken and Grill

Texas Fried Chicken & Pizza

Bbq Chicken Food Truck

Chicken Express

The Chicken Wing District.

Best Chicken

Harvell's Chicken On A Stick

Natural Chicken Grill

Chickie Wah Wah

 

 

Natural Chicken Grill

Chickpeas Vegetarian

Chicken Al Mattone - Frisco

Lebanese Chicken - Hamtramck

Mexican Grilled Chicken

The Chubby Chickpea - Closed

Chicken Cabaret

Chickpeas Vegetarian

Chick-fil-a of Dover - Closed

Feisty Chicken Grill

Chick-o-pea's

Chickpea Mediterranean Grill

Natural Chicken Grill

Fox's Pizza & Chester Fried Chicken

Chick-o-pea's

Vitamin Chick

Chickpeas - Closed

Aunt Norma's Fried Chicken & Title Loans

California Chicken Cafe

Best Chicken

Harvell's Chicken On A Stick

The Chicken Wing District.

Kentucky Fried Chicken

New York Pizza and Chicken

Chicken and Guns

 

"garbage in, garbage out."

(These datasets need to be inspectable.  This is a challenge as they get big.)

See also survey in "Procedural Content Generation via Machine Learning (PCGML)" (link)

We need to be careful what we are teaching these cultural objects...

The data we save and analyse, the tools we create are political artifacts.  Who creates them is political.

So are the stories we tell.

Who gets to tell stories? 

and with what data.

Ask yourself...

  • What story is investigated in the first place? Who decides?
  • What viewpoint is presented?
  • Who is empowered to speak and be heard?
    • trained in the tools,
    • their work distributed,
    • are paid for doing it.

(Computer science is

 not split out. It would

be good to track this.)

Who is recording history?

Surveys have indicated that between 8.5% and 16% of Wikipedia editors are female.[4][5][6] Consequently, Wikipedia has been criticized by some academics and journalists for having primarily male contributors,[7][8][9] and for having fewer and less extensive articles about women or topics important to women. The New York Times pointed out that Wikipedia's female participation rate may be in line with other "public thought-leadership forums".

Another critique of Wikipedia's approach, from a 2014 Guardian editorial, is that it [sic] has difficulty making judgments about "what matters". To illustrate this point they noted that the page listing pornographic actresses was better organized than the page listing women writers.

Who gets to be the subject of the story is an immensely political question, and feminism has given us a host of books that shift the focus from the original protagonist—from Jane Eyre to Mr. Rochester’s Caribbean first wife, from Dorothy to the Wicked Witch, and so forth. But in the news and political life, we’re still struggling over whose story it is, who matters, and who our compassion and interest should be directed at.

-Rebecca Solnit, "Whose Story (and Country) Is This?" (2018)

This documentary was made by 2 women, Trish Adlesic & Geeta Gandbhir.

Adlesic produced the documentary with Mariska Hargitay, who played Olivia Benson on Law & Order: Special Victims Unit, a character who advocated for survivors and had herself been sexually assaulted. Hargitay is an advocate for sexual assault victims in part because people would write to her about their abuse stories after seeing Law & Order. “At first it was a few, then it was more, then it was hundreds, then it was thousands"

Indignation, passion or even just interests of difference...

in

2013:

I remember this,

and I thought

"that's fun and different!"

and cute.

Lyzi is a software engineer

at Mapbox now (afaik).

Inviting stories: Different kinds of diversity to consider.

Genders & Race

Age

Interdisciplinarity (Engineering + Social Sciences + Design + Humanities....)

Occupational History & Training

Accessability

Cultural

Linguistic

Software background

...

 

 

Thanks!

@arnicas

arnicas@gmail.com

data-vis-jobs-fr@googlegroups.com

data-vis-jobs@googlegroups.com

P.S. Job postings: if you are trying to hire data vis help - post here: