Build Your First Software: A Historian's Introduction to Writing Code

Abstract

A beginner-friendly session on writing software for historical research, taught from a research software engineering (RSE) perspective. Jeri Wieringa shows how large language model (LLM) coding tools such as Claude change what it takes to build a research tool, using a live project that turns 19th-century statistical reports into a spreadsheet. The demonstration focuses on project folders and written plans, with version control handled in Git. She closes by showing how the same files can move into other tools such as VS Code.

Chapters

1. 0:00 Welcome and Jeri Wieringa on research software engineering at Princeton
2. 3:18 Three levels of using AI like a software engineer and what the room has built
3. 8:51 Claude Cowork and the harness software around a language model
4. 15:46 Level one: where computation fits in historical research
5. 19:58 Project folders and context for a language model
6. 25:08 Live demo: a Claude project for 19th-century statistical reports
7. 31:04 Level two: asking for a plan before any code is written
8. 40:40 Saving the work as durable, repeatable Python scripts
9. 48:48 Version control with Git
10. 51:44 Memory and context limits measured in tokens
11. 56:52 Level three: working outside Claude in VS Code
12. 1:01:43 Questions on Zotero and the Model Context Protocol (MCP)

Transcript

Transcript not published at the speaker's request.