๐ฟA shared experience of affirmation
๐A generated poem made from class notes, a loving reminder of what we talked about, shared and learned.
๐ฑThe rest of this page contains reuminations and ramblings that ultimately led to the creative output contained in the links above.
Learning to code has taught me that learning is an emotional process.
Coding gives a tangible feeling of satisfaction when you write something, run it and something happens.
I've also been thinking a lot about being a 'coder', a 'programmer', how and why that became an identity.
And whether I want that as an identity.
Knowledge and skill as an identity has its roots embedded in our capitalist societal structure.
From birth parents try to identify what their child is 'good at', and all through our years in school, we are told we are
'good at xxx'.
Don't even get me started on 'gifted'.
'Good at x' becomes an identity. At the same time, perhaps out of rebellion, 'not an x person' also becomes a valid identity.
These identities that we take on become emotional guides to what we should be able to do, and what we can take perverse pride in not understanding.
As I feel more and more secure in my understanding of (very few!) coding languages and my ability to Google things,
I'm not sure I want to take on another identity right now. I think it's OK to be a person that can do, and enjoys doing some code.
I have been thinking about this project and trying to clarify the concept -- Who is it for? What do I want it to say? How can it be clearer / more self contained?
I read an article recently about reciprocity, about not reading as consumption and it resonated with me to try and make something that by encouraging reciprocity also breaks down the distinction of programmers on one side and users on the other.
Somehow that tied to an idea I had to make this project for my fellow classmates -- as an affirmation that we are all coders and programmers. That it's more about a frame of mind than being able to type code from scratch.
I also wanted to make it more self contained so I've decided to put it all on on webpage: I will try to make a generated affirming poem from class notes, a text input box that people can write in and the fractal tree will grow from the input people give it.
I want to create a website than makes people feel heard / seen / cared for in some way. The internet, and technology more broadly, can feel like a somewhat unsympathetic black hole -- users give their data, attention and money for services which are often not provided with care.
I find it difficult to convey care with machine learning alone -- by nature ML algorithms take in any text, image or object just as raw materials with no context or (human) comprehension. What it does with the input feels alien and can be fantastic but not warm.
So I've decided on a combination of twitter bot and website that will take tweets and either respond or use the tweet to 'grow' a tree. The twitter account will ask people to either share something positive - which will case the fractal tree to grow on a website; or they can ask the bot to say something nice to them if they're feeling a bit blue.
To help something to grow feels rewarding. Hopefully this also applies to a virtual tree on the internet. I chose a fractal tree because it's generated by code but through our human eyes it turns into a tree. I'll attempt to compose with care an algorithm that can craft generated responses that feel warm, even as I plan to make it known that it's a machine tweeting back. Perhaps something that can break down parts of speech and use relevant word(s) in its response. The tree will grow with each tweet that's sent to it. Eventually it may grow too big, at which point new tweets will cause it to glow or change colour.
Below is a tree made in p5 for illustration. You can click your mouse anywhere on this page to make it grow more branches, and press any key to make it grow leaves.
Use Python and tweepy to gather tweets from account
Tweets analyzed and stored in either text files or csv -- Python to write to file for storage
Final analysis method tbd, but can analyze by parts of speech and sentiment
Generation needs to be experimented with - textgenrnn? markovify? mad libs?
Need a sample dataset to start with
Python to detect if tweet says "tell me something nice", in which case use tweepy to reply with generated tweet
Script to be stored on vm to run x times daily
Webpage on load to check number of tweets received by account
I hope this can be done by Javascript - either to check lenght of a csv or for Python script to make a counter file for it which just holds a number that is updated
The number will determine how many iterations of generation the tree goes through ie how big the tree is
Tree is made in p5
Experiment with background, sound(?), simple animations