With 16 bots and hundreds of other coveted projects to his name, Beau Gunderson is a prolific developer and valued contributor to the open source community, responsible for adding thousands of unique artworks and text snippets to the Twittersphere. And Twitter really is the ideal gallery for this form of expression; ideal for audiences because it allows them to become immersed in the bot’s stream of consciousness, and ideal for developers as an easily-accessible sandbox in which to store and evaluate new ideas quickly.
Since the earliest days of Secret Cave, we have been obsessed with the art of bots and the motivations of their creators. Here, we take a closer look at three of Gunderson’s inventions which intrigue and inspire in equal measures.
For those who watch films with sound effect captions turned on, @sounds_distant‘s output will be instantly familiar. For this bot, Gunderson scraped data from the OpenSubtitles2018 database — a set of almost four million subtitle files — and filtered it to include only “soft”-sounding lines. This meant including lines that match terms like “muted“, “distant”, and “soundless“, and excluding lines with “explode“, “cannon“, and “gunshot”. You can see the filters in the bot’s filter.js file along with the rest of the source code here.
indistinct; ashtray clanks
— in the distance (@sounds_distant) June 10, 2018
swords clashing in distance
— in the distance (@sounds_distant) May 30, 2018
telephone ringing in distance
— in the distance (@sounds_distant) May 26, 2018
@threedesert is one of the more striking generative art bots on Twitter. Combining skeletal renderings of commonplace objects with contrasting block colors, its output is often an intriguing visual puzzle. The objects clash to create new shapes and textures as they intersect which gives it a sort of 3D cubist feel.
The bot uses Michael Fogleman’s ln and gg Go libraries to render its images, but exactly where it gets the blueprints for such accurate and complex objects is a mystery to me as a novice Go reader. Check out the source here and see if you can decode it.
— threedesert (@threedesert) 11 June 2018
— threedesert (@threedesert) 9 June 2018
— threedesert (@threedesert) 6 June 2018
L-systems were originally developed to model organic growth (similar to and compatible with Conway’s Game of Life). As you’d expect from a process that takes a random seed, the result is often chaotic. However — like in nature itself — the results are occasionally clear, symmetrical, and geometrically perfect.
— lindenmoji (@lindenmoji) 31 December 2015
— lindenmoji (@lindenmoji) 11 June 2018
— lindenmoji (@lindenmoji) 10 June 2018