Final Project: NaNoGenMo Update

About

For my final project I built off of my NaNoGenMo project (which you can read more about in my previous blog post here).

My goal for the final was to make my NaNoGenMo visually more interesting, improve its structure and coherence. I did so by doing the following:

1. Added chapters, one for each month of the year

2. Assigned the objects into months using the correct number of calendar days

3. Use these month groupings to generate images (using AttnGAN in Runway)

Reflection

Because of the difficult nature of narration and infinite ways novels can be computationally produced, I’m not sure if I will ever feel done with this project. I have really enjoyed working on it and learned a lot in the process (technically and conceptually). I found it to be an interesting exercise to try to break down a novel this way and think about applying the concepts we had learned in earlier on in the semester using other tools. I also found it a refreshing way to make a project around questions I have about human-computer collaboration and attitudes/approaches towards technology, algorithms and decision making. Although this project was meant to be subtly critical, it was also a curious project in terms of seeking to know what we have unknowingly taught “computers” and one that I felt had a positive, warm and nurturing attitude towards “computers” - similar to Karl Ove Knausgard’s sentiment in his letter to his daughter.

Links

Here is a link to my presentation slides.

You can see the result on Github here.

My python code is also on Github here.

The novel formatted as a webpage is here.

NaNoGenMo

Inspired by the book Autumn by Karl Ove Knausgard (KOK) in which he writes about a different object each day (for his unborn daughter in order to teach her about the world), I decided to generate a novel using this same structure but from the computer’s perspective. With this exercise, I hoped to explore the following questions: what have computers learned about the world? And what would computers tell us about the world if we asked them about it in 2019?

I decided to use the top 100 books from Project Gutenberg from the last 30 days as a proxy for all the things we have collectively been teaching computers (“computers” here is also a proxy for really what should more specific questions about individual algorithms or computational systems).

I took the top 365 nouns from these books, one for each day of the year. Then I pulled out all the sentences that included each of these words to create a “corpus” for each object. For each object I then used markovify to generate 100-200 words “about” that object (5 sentences).

KOK starts his entries about objects often with technical descriptions (followed by more narrative-type text), so I looked into adding wikipedia descriptions using the wikipedia python library. I had hoped that I would be able to easily add the first part of the, but got many errors because there was not an exact match for a wikipedia entry. I tried to instead choose the next wikipedia entry, but often this would throw disambiguation errors too or lead to a very random description (about a song for example). With this method of going down the line. For now I left these cases (32/365) blank if there was not a good match and I will plant to fix these in the next phase (I think I may have to fix them by hand unless I can think of a better way to do it computationally). For the ones that matched, I included the first three sentences of the wikipedia entry.

The output doesn’t make much sense, but I do find the juxtaposition of the technical text with the generated text interesting at times when they end up connecting or reflecting each other - in particular seeing which elements carry over. I also find it a fun exercise to try to guess which novels the generated text is pulling most from and thinking about why that is. I have kept the output in the order of most common to least common nouns, which gives another kind of insight into the type of source text this uses - I think it says a lot that the most common noun is “man”. My hope with this project is that these kinds of questions can prompt deeper thinking into what we are “teaching” computers and the care and intention we put into that act.

Here are some links:

Github repository which has the following:

• Python Code

• PDF version

• Plain Text version

NaNoGenMo Official submission!!

Week 7 + 8: Dialogue and Character with Ren'Py

For this week’s assignment to create a character dialogue or chat bot I returned to a book I’ve been wanting to work with for some time. Last year over dinner one night, my aunt’s friend (Kim) told me about an experience she had on the WELL (the early BBS “chat room/social network” system) - she was too shy and new to computers to participate in the conversation, but observed as a topic called Local Bug Report unfolded. It started with a user with the name mandel (Tom Mandel) who posted about the flu he had that was “going around the office.” As time passed and the discussion continued, he visited the doctor and learned that he had lung cancer. The topic turned to a report on his health status and a sort of support group as he was going through treatment. I am especially interested by this in terms of internet history (and specifically the development of communities online) and as possibly the first death online. My aunt’s friend said that when she saw this happen and the community’s reaction, she understood what the internet was about - or would be about.

Kim published the transcript from the chat in a book called @Heaven (with permission from the users). I read it last year and it was unlike any other book I have read because it wasn’t really a book or novel with a traditional narrative. There was narrative present in it, but it was driven by a natural progression and dialogue facilitated through the medium of a chat room. Most of it was very mundane but it was punctured by unpredictable poignant, moving, and intense moments.

There is a moment before mandel dies that he speculates about someone making a chatbot version of him in the future so that he could “live on” virtually. Since reading that I was interested in trying to complete that kind of project and this assignment seemed like a good opportunity to explore this idea.

I tried to find a version online I could either download or copy and paste lines from, but could not find a format that would work. Since I didn’t have enough material, instead of making a chat bot, I decided to make an interactive recounting of this piece of history to think about what this book might look like as an interactive conversational narrative similar to Butterfly Soup.

To do this, I ended up typing out sections that I wanted to include in order to create a semi-branching narrative as a prototype for what this experience could be like. As I was doing this, it actually felt important to type out the text for this project and I was glad I was doing it rather than copying and pasting or using a text corpus version of the book - it made me feel more of a connection to the people who were participating in this chat and better able to empathize with their experience.

I included some options of responses to different parts of the chat based on real chat responses in the transcript (“beams” for example was a word used a lot in the chat to send good thoughts/vibes/prayers for example). These do lead to a few different branches of the narrative at first, but all come back to and in the end the options do not change the path of the narrative.

The only other character I included was a post from “nana” who was mandel’s wife (they married shortly before he died).

And I have a final room, which is the final discussion topic mandel started called “My Turn” to say goodbye to his community:

One challenge was that I hadn’t realized that may of the posts are quite long and Ren’py doesn’t seem to handle this well. I split up the text into smaller sentences, but it became odd when I started making the menu options large chunks of text. Perhaps a better solution would have been to have single word menu options and then show in the bottom what the full response was.

I also wasn’t what to do about visuals. I don’t think I’d want to include pictures of actual people - maybe if I were to continue this I’d add abstract visuals or just a picture of the computer the character is typing at.

You can play the Ren’Py version here: https://lydiajessup.github.io/mandelbot/index

Week 5 + 6: Setting, Space, Props

This week our assignment was to recreate a scene or location in Inform7. I decided to recreate a scene from Axolotl, a short story by Julio Cortazar (Spanish here and English here). I love the magical realism writers from El Boom Latinoamericano and the writings from these authors touch on a lot of the themes we have been talking about with time, nonlinearity, branching narratives and hypertext. I am also developing this story in VR for another class, so was interested to see story can be broken down into smaller interactions and steps that guide the reader/immersant.

In this story, a man visits an aquarium in France and becomes obsessed with the axolotls. He continues to visit them, staring into their eyes and contemplating what their thoughts might be. Eventually he becomes an axolotl. 

I was interested to see how a stripped down medium of interactive text could guide the reader/player through the actions of this man. Initially I wanted to start with the man biking the streets of Paris, but this became too many steps to start with, so I started inside the Zoo. At first I was going to have one “room” and have all the animal exhibits in the same room and the user could approach, but I found it easier to keep track of and more interesting to create multiple rooms to explore.

I wanted to make an animal class and have the actions you could take on the animal depend on the kind of animal, but found this way of coding a bit confusing and could not resolve the errors I was getting when trying to write if statements for these conditions.

In the end I decided to create a boolean to indicate whether the animal is an axolotl or not in order to differentiate the actions you can take on it. It seems a bit inefficient the way I have it set up though. I’d also like to “unlock” certain actions once others have been performed and again am doing this a couple times with booleans, but it seems like there is probably a better way. For example, right now you can “stare”, “tap” and “continue staring” at the axolotl without approaching it first, but I’d like to make approaching a condition for these other actions. I’d also like “continue staring” to only be possible after you have tried staring, maybe even a few times.

You can play the game here: https://lydiajessup.github.io/inform-sketch/play.html

Try:

• Greet

• Contemplate

• Approach

• Stare

• Tap 

• Continue Staring

Week 3 + 4: Hypertext & Twine

I have recently been researching the history of hypertext in the context of the development of the internet, so was excited to use this assignment to explore this research using new methods. This research started when this summer my friend and classmate Emma Rae Norton and I researched and designed a Feminist History of the Internet Walking Tour for ITP Camp. I am now working on converting part of it into an Adjacent article, focusing on the 1980’s and 1990’s in lower Manhattan. The writing process has been challenging partly because I hold a lot of the connections between people and events in my head and I find it hard to write them down into a more linear historical narrative. I once tried to drawing some out on a huge piece of paper and Emma and I started to hand code a website to visualize and experience them in a different format.

For this assignment I was interested in using Twine to take the reader on this tour in an interactive and networked format. I was also curious as to whether this would help me discover new connections between places and people or more clearly visualize the strongest nodes. 

I had planned to go through my Adjacent article draft section by section and add snippets of the text to nodes, but I found myself jumping around. I decided to let this happen and wasn’t too surprised that this was how my brain thought about and processed all of these dates and facts. I was also interested in what the most commonly visited nodes or links would be, so added in the code to track how many visits each page gets.

As I was creating the nodes in Twine, I also kept trying new groupings to try to find the ones that looked better visually, but eventually gave up on making it look “clean.”  Some groupings emerged, but one of the main takeaways was that the connections are more complex than I had previously realized. There seem to be infinite connections I could add. New ideas for how to split up the nodes also kept popping up in my mind as I was constructing this. I enjoyed hiding little “surprises” that were harder to get to with only one link in (like Jaime Levy’s rap) and wondered if people would be able to get to them.

You can see this online version of the tour here.

If I were to continue developing this, I’d like to add photos and create pages for some of the other connections for common phrases (such as media studies) and years.

Week 2 + Presentation

Presentation

This was my week to give a class presentation and I was assigned the topic of alternate reality games. Below is my presentation.

Here are links to my sources:

Alternate Reality Game, Wikipedia

• I Love Bees, Wikipedia

• I Love Bees, 42 Entertainment

• The Beast, Wikipedia

• Cracked, 5 most insane alternate reality games

• Argn.com

• ARG Reddit

• UnFiction.com

• The Year in Alternate Reality Games, Wired

• Alternate Reality Games Could Still Take Over the World and Your Life, Atlas Obscura

• Alternate Reality Games Blur Reality and Fiction, Reuters

Playing Games

My Body

This piece uses hypertext to explore different parts of the author’s body. There is an entry page, but I soon got lost moving through the hypertext links, sometimes ending up on the same page and sometimes wondering how to get back. This also allows the reader to start to re-arrange the piece - for example, reading one paragraph from the first page, clicking on the hypertext link and reading that passage, and then going back to finish the first page.

In terms of decisions that the reader has to make, they have to decide where to click first and how to move over her body, which part to move to next through hypertext links or if they want to try to return to the first page. Other questions that that came to mind was, should I read all of the text on each page or jump around and eventually circle back enough that you read it all?

Hypertext works well here - it adds understanding of how the authors sees parts of her body as connected, like a map or a mental map of a space. In some ways though it was hard for me to finish reading each passage because I wanted to see what the other body parts were she was referencing. I found myself clicking back and forth, sometimes being surprised sometimes thinking “oh that makes sense” and more often than not comparing to my experience with my body.

Depression Quest

This narrative has branches but still keeps a linear structure that the reader can go back to, unlike My Body in which the reader can choose their own path and enjoy feeling lost. This structure is very clear to the reader with the presence of “back” and “next” buttons which don’t exist in My Body. Within this structure, it gives the reader some choices that lead to different branches but all continue linearly moving the story forward.

There are points where the reader has to make decisions to move to different branches. In some cases, choices have been taken away, in this story indicating what someone who is not depressed would do. Within each branch, the reader can choose to get more information about a certain area of that path or move forward quickly without reading deeper.

Hypertext feels appropriate here because the story’s goal is to put the reader in the position of a person who is depressed and have to make hard decisions, even though many of the options feel small in the moment the hypertext function helps the reader understand how they have different outcomes and lead to new choices.

A Dictionary of the Revolution

In this dictionary hypertext us used to visually show the reader the “strength” of the connections between words. Instead of linearly reading through the dictionary, you can click through in the order you choose and explore connections in a manner that feels more natural and dynamic.

The reader has to decide how to navigate through the dictionary: do they read through in alphabetical order, visually noting the connections? Or click through to read about the strongest connections? Or choose what looks to be the weakest connections? Or do they click through at random?

Hypertext feels appropriate here and to me as a reader feels more intuitive than trying to read a list or dictionary in a linear way. Sometimes the number of choices feels overwhelming, but that is also reflective of reality. As a reader, I really got a sense of how complicated this environment was that I think I wouldn’t have understood if I had simply read the definitions for a list of terms.

Similar to 80 Days, this experience feels less like hypertext (in a traditional sense on the internet with links) than My Body and Depression Quest and more like an “interactive web page” or data visualization. This story is most similar to My Body because there is no prescribed way of navigating the site and it’s clear how each page (or term) references and is connected to the others - the authors have laid this out clearly and a big part of the “journey” is exploring, understanding and thinking about these connections..

80 Days

This game uses hypertext to make decisions in the form of answering questions and taking new actions. These lead to additional narratives and new options for decisions and actions you can take.

The player has to decide how to answer questions and what actions to take next. These often feel like small differences and sometimes the player get forced in into one path after-all (for example, pretending to lose the tickets and being forced to say that you forgot to buy them).

Because this is a game, the branching narrative feels appropriate here. Even though it’s a linear journey the player still feels as though they have a lot of choice over which direction to go and what to do. Perhaps it was the phone interface, but at first it wasn’t clear to me what action I “should” take or if one was necessary at all. While it’s not clear how much the choices change the outcome, the small choices of how to respond did make me feel as though I had my own personality, rather than being forced into a character chosen for me.

Compared to the other stories, this feels less like hypertext because of the game elements and most similar to Depression Quest in its structure (but a much more complicated version).

Sketch 1: Tracery

When I was a kid, the book In a Dark, Dark Room and Other Scary Stories was very popular at my elementary school. One of the scariest stories to me was The Green Ribbon about a girl who wears a ribbon around her neck and when it is untied her head falls off. As a child this was one of the scariest stories to me (you can see the buzzfeed reaction here, proof that I wasn’t the only one who had nightmares about this for years).

A couple years ago I read Her Body and Other Parties which revisits this story and re-tells it in a short story called The Husband Stitch. 

As the name hints, I think that this story can still be classified as horror, but I now see the story as scary from girl/woman’s point of view instead of from the boy/man’s. I now see that the conflict and violence is drawn out over the course of the entire narrative. I love this story because I feel it captures a part of my experience of growing up as a girl and being a woman in our society that I have never quite been able to put into words. 

For this assignment, I decided to revisit this short story to understand what Carmen Maria Machado changed about the discourse that showed the story in a new light for me.

For reference, here is a short version of the story, similar to what I read as a kid.

I was curious about the following: how does the story change if the ribbon is tied around different body parts? How much does the ending (the exchange between the husband and the woman) change the rest of the story? Does it matter what kind of man the husband is (good, bad, etc) or what kind of woman the wife is (beautiful, plain, odd)? Does the story change if the woman has a daughter instead of a son? 

I wanted to retain the original story I read as a child, but in order to add the other variations I wasn’t exactly able to do this, although some versions are close. The changes in the story are subtle, but I think do achieve telling the story with different discourse - a head falling off is very different than a finger falling off, and a “bad” man asking to untie the ribbon is different than a “good” man asking. The core story remains the same: this is a normal and unremarkable love story (meaning a heterosexual relationship between two cis people) and as time passes they meet, fall in love, get married and have a child. Throughout, the boy/husband/man asks over and over about the one secret the girl/woman/wife has: the ribbon tied around part of her body. Finally, at the end of the story, he unties it.

Below are a few different versions generated by tracery.

You can see my code here: https://editor.p5js.org/lpj234@nyu.edu/sketches/1q9s0BcXSV