This website is an attempt at building out my personal “forever home” on the Internet. Inspired by Gwern.net, it contains my bookmarks, videos and papers I’ve read and/or collected since the 1st of January 2024, along with my writings and other goodies.

The website was built with Hakyll, a static site generator written in Haskell. The classification system for resources is leveraging OpenAI’s o3 and is written in Rust as a CLI and TUI. The process looks somewhat like this:

• I have a single sqlite database
• The Rust CLI is in charge of contacting various APIs and syncing bookmarks, videos, and papers to the database first.
• The Rust CLI is also responsible for prompting o3 to classify the resources onto the tag tree, and has various functionalities to manage the health of the tag tree such as pruning, merging, and splitting tags.
• Hakyll (using Pandoc) will query the DB and generate the HTML tags (along with some processing for dates, bibliographies, truncating various things for formatting, etc)
• a CF worker is currently used to “host” these static HTML files (the domain was purchased through Porkbun)

Over time, I intend to build out a few more things such as classifying my web history to capture “research trails” e.g:

Research Trail: Category Theory Deep Dive
    Period: 2024-03-15 to 2024-04-22
    Intensity: High (85th percentile)

    Trail Progression:
    1. General Interest (Wikipedia: Category Theory)
    2. Exploration Phase (nLab, various blog posts)
    3. Paper Reading (ArXiv: 15 papers on CT)
    4. Applied Interest (Haskell libraries, type theory)
    5. Connection Phase (Links to algebraic topology)

    Key Domains: http://arxiv.org, http://ncatlab.org, http://bartoszmilewski.com
    Peak Activity: 2024-03-28 (47 related visits)
    Total Investment: 28.5 hours

along timeline visualizations, automatically publishing my Obsidian thoughts etc.

Much of the point of this website is personal: I am interested in durably saving my various “arcs” and adventures in figuring out what interests me the most around and letting grow over decades. I am archiving a lot of my data over time, including every web page I browse, every video I watched, any software I use that has any sort of scrappable or extractable data (such as a sqlite file for browsers storing history and bookmarks), etc.

You can find all of the resources exportable as JSON below:

• Bookmarks
• Videos
• Papers