Site Navigation

Edit on GitHub


Handwriting History Project

Introduced by: Dave Kaminski
Depth: Medium thread

Dave's passion project: researching, cataloging, and publishing historical American handwriting samples. It's the reason he started using Claude Code in the first place, and it's the domain where both his best results and worst failures occurred.

What He's Built

The Research Challenge

The data is deeply nested and full of exceptions:

Dave: "My whole research is based on not just the core, but on the outliers... and then tracking those back and forth."

Example: Palmer Method was an American handwriting style, supposedly ending around the 1930s. But Dave has a 1943 letter from a ship captain written in Spencerian — a style that "ended" in 1897. Tracking these outliers requires classification at multiple levels: style, date, geography, certainty level.

Dave wants to classify items as: certain early, possible early, possible late, certain late.

Claude's Role

Claude built the website, the map, and the thumbnail viewer. The structured data work went well. The unstructured research work — trying to get Claude to manage nested classifications, track outliers, and maintain a coherent taxonomy — is where things fell apart.

Dave: "One of the tricks is to really have your stuff originally in a structured data that's actually correct. And if you can do that, then this thing will just do anything for you."

Pete's Suggestion

Use a project-based vault structure in Obsidian, with sub-projects for specific tasks. Pete created a Palmer vault live during the call, demonstrating how to break the research into manageable pieces. See Putting Claude on Rails.

Related


Pages that link to this page