My latest AI gardener apprentice finds pairs of notes that aren't explicitly linked, but maybe ought to be. This project was inspired by the human subconscious, which creates, cements, and removes neural connections during sleep.

Like a sleeping brain, my python script runs nightly, scouring my note garden for related notes. The end result is a list of the most similar notes pairs in the garden, based on semantic similarity. I found the results to be illuminating, and a real-time version with solid UX would be a central feature of my system for thought.

One benefit of keeping a long term note garden is that you can have conversations with your past self. You wake up a slightly different person every morning. Aggregated over years and decades, you slowly become a very different person. But if you've been taking notes, a crystallized past self is still around!

Unfortunately, your past self kinda sucked at note taking. He wrote too verbosely. He was not familiar with note hygiene like putting the Bottom Line Up Front. Worst of all, he picked the least descriptive note names, like "3D Automaton", "Geo Games", and "Run". If he was still around, you would tell him to read How to Take Smart Notes by Sonke Ahrens, but your old self is long gone.

What if we could summarize overlong, unstructured, poorly named notes automatically, into terse but precisely descriptive golden nuggets?

Tools for Thought are all the rage these days. Unfortunately, most of these tools assume one corpus per tool, and the interoperability story is very poor. At the same time, it's pretty clear to me that no single app can serve as my second brain. What I need instead is a System for Thought that supports many different tools working seamlessly together.

The solution I have arrived at is a cloud-synchronized directory of plaintext files and images. Here's a visual summary of what's going on in my system for thought today:

This system has two key advantages:

1. Tool choice: I can choose my favorite among multiple existing tools
2. Hackability: I can build my own tools to support specific needs

The ancient Sumerians invented writing sometime around 3500 BCE. But how did writing get refined? How did it spread outside of Sumer? In general, tracking diffusion of ancient technology is hard. For tracking the spread of literacy, however, here's an interesting idea. Innumerate people may fudge their self-reported age to round or auspicious numbers. Individually, this leads to terrible earworms. In aggregate, this error is called age heaping and may be a decent proxy for literacy. In this post, I dig into the Province of Canada's 1852 census, scraped from automatedgenealogy.com. To whet your appetite, just look at these beautiful age heaps:

Splitting the census data by demographics and calculating the ABCC Index on each, what can we infer about literacy in 19th Century Canada?