pkb

Our goal, simply stated, is a note system that levels off complexity through always-improving organization.

Background

Scientific collaboration. How can we combine ideas whose complexities are beyond individual comprehension? Science is beginning to develop and apply systems for large-scale collaboration (Reinventing Discovery). We're investigating how this plays out with the individual partipant grappling with the space of ideas. From our perspective, the classic notebook is still the basic metaphor for this task, but we can amplify its presentation, interactive capabilities, and collaboration. Work with the federated wiki (see Mike Caufield) is one inspiration for this digital tool.

Compressive file systems. In his Strange Loop 2014 talk, Joe Armstrong described a compressive file system that would reduce the space taken by files with compressive similarity. An easy way to incorporate this principle is to use content-addressable storage for data that we store immutably, such as quote from websites.

For a "compressive note system", we can go further: 1) Compressive similarity can be extended from bytes to concepts. 2) Our notion of complexity is not measured in disk space but in human computation. For example, a note that is unintelligbly compressed to 50% that must be extracted and read in full is measured differently from one that is compressed to a half-length summary that I can scan instead of reading the full note. (The former has a compression of 0% and the latter ~50%.)

Sensemaking. Only an organized note system has a chance of leveling off in complexity. While organization can be described as a process leading to an elegant external representation of concepts and relations, sensemaking is the other side of the same coin: a process leading to an elegant internal representation of concepts and relations. One cannot form a valid external representation of concepts and relations with out having made sense of them except by luck. However, it is with an manipulable external representation that internalized is most likely to happen. (What do we take from this? Are we inspired by evolutionary processes or other optimization techniques?)

Usage

Check out the demo on pkb.herokuapp.com. It currently saves data to your browser's local storage.

To setup locally, git clone https://github.com/ryanmuller/pkb.git then in the pkb directory:

1. Install requirements: pip install -r requirements.txt
2. Run the server: python app.py 8080
3. Navigate to http://localhost:8080

File storage version (experimental!!!)

An experimental branch of pkb can read from and write to a directory of Markdown-formatted files as the data backend. Be very careful using this branch because it is not well tested. NOTE: Currently only the first block in a page saved at all!

1. In the pkb directory checkout the file storage branch: git checkout file-storage

2. Modify DATA_PATH, PAGES_DIR, and PAGE_EXT to match your system.

3. Make sure that you server process has permission to write to these files. For example, my files are writeable by the _www user, so I run the app as that user

   sudo -u _www python app.py 8080
   

   Make sure the static files are all readable too (since the owner may now be different):

   sudo chmod -R a+r static/
   

Roadmap

Flexible storage. We use content-addressable storage to facilitate import to and distribution of your knowledge base.

Ubquitous capture. We provide web capture tools and support import from other APIs and RSS feeds. We also define a standard format (similar to Jekyll front-matter) to specify sources.

Powerful sensemaking and organization. We offer UI affordances and automation tools that help with the process of organizing and making sense of your information.

Web and mobile publishing. In addition to running a server on your CAS, we allow export of static HTML to be easily uploaded to any host.

Next steps

See waffle.io/ryanmuller/pkb