The medium is the message.

Diverge at any and every letter.

This project is mostly just an idea so far.

But check out the beginnings.

You'll be able to forge a path by typing, or follow suggested paths via mouse/pen/touch/whatever (or Tab on the keyboard).

Currently you can only use the keyboard, and to follow a suggested path, you have to type it out.

Similar paths that don't quite match could be lined up "magnetically"

Here's some crude ASCII-art to try to communicate the idea:

The quick fox jumped at the opportunity
The qui   fox jumps     the
       c n         over     lazy dog
      k   w
       bro

Providers can look at the path you're on and the cursor position and suggest new paths.

The basic provider returns paths from a database of strings entered by users.

A Markov text chain or sentence autocomplete provider would return paths with text inserted at the cursor, whereas a paraphrasing provider (possibly using a neural network?) could return strings with infix replacements, making for paths that diverge and rejoin the current path.

Do you value rhyme, alliteration, assonance? Are fifthglyphs vile & evil, or good, top notch stuff?

Arbitrary writing constraints and softer nudges can be interesting instruments of creativity.

Evaluators can be combined and weighted.

i think it could be a really cool unique social experience

branching off of each other's writing

sharing in the constraints for creative writing with a consistency across the collaboration

a natural medium for nonlinear stories

I don't know what would work best for the granularity and structure, but it could have communities with independent moderation and rules, similar to Reddit, and pages with seed texts... and then maybe special rules could be applied to certain pages, possibly based on tags... and maybe some tags could only be set by moderators and some could be set by anyone...

A simpler model could be that there are just pages with seed text and rules set initially by a user.

Either way, there'd be links, of course.

An alternate (maybe not exclusive) direction for this project to take could be towards longer-form writing like articles and even books. In that case it should work inline, and it would be more focused on editing, rather than pure exploration. I think this is less interesting for now.

• a different textploration idea: Intimate Codex
• the LibraryOfBabel subreddit
• the LibraryOfBabelCollaboration - a google doc you're free to edit
• Your World of Text and Our World of Text
• Dasher, a text input method where you zoom and drill down into letters, weighted by how likely it thinks you are to want to use them from context
• rnn-writer, a sentence autocompleter, which could be integrated into this project, via their torch-rnn-server, or even their hosted text.bargains
• gpt-2, a newer text autocompleter

I'm using Create React App for the webpack + dev server setup, even tho i'm not using React.

• Clone the repo
• Install dependencies with yarn install

In the project directory, you can run:

Runs the app in the development mode.
Open http://localhost:3000 to view it in the browser.

The page will reload if you make edits.
You will also see any lint errors in the console.

Launches the test runner in the interactive watch mode.
See the docs about running tests for more information.

Builds the app for production to the build folder.

The build is minified and the filenames include the hashes.

Builds, and then pushes to gh-pages

MIT licensed. See LICENSE.txt for details.