About the Larabie/Lavigne Genealogy Project 🌳📦🤖
Hello! Raymond Larabie here 👋
This project started with several dusty boxes in my dad Roger’s basement. Inside were decades of family material: daisy-wheel printouts from the late 1980s and early 1990s 🖨️, handwritten notes ✍️, church records ⛪, newspaper clippings 📰, newsletters, family binders, photo captions 📸, oral-history fragments, fading photocopies, and all kinds of mysterious paperwork held together by staples, tape, and pure optimism.
Some pages were beautifully organized. Others looked like they survived a flood, a raccoon attack, and three generations of coffee spills ☕🦝
Instead of entering everything manually into traditional genealogy software, I decided to try something unusual: use AI to help turn the archive into structured text and a navigable family graph 🤖🌳
How it works ⚙️
Every scanned document was converted into tagged text. Dates were normalized into standard ISO format, French material was usually translated into English ➡️, and conflicting information was intentionally preserved rather than silently “fixed.” Different source types were also assigned different trust levels.
That means:
- a typed obituary might be considered fairly reliable ✅
- a half-legible photocopy from 1989 might be “probably right” 🤔
- a whispered family rumor from somebody’s aunt at a funeral might still get preserved because honestly those stories are often the interesting ones 👀
The goal is not to create a perfect legal registry. The goal is to build a living family memory map 🧠
The weirdness is intentional 🌀
Family history gets messy fast.
People appear under different spellings, nicknames, married surnames, missing accents, OCR mistakes, or names remembered differently by different relatives. Some branches are richly documented; others are mostly reconstructed from fragments and detective work 🕵️
Rather than hiding uncertainty, this project tries to preserve it.
That’s why you’ll sometimes see floating “loose thread” people drifting around the outside of the graph 🧵
Those may represent:
- possible relatives
- duplicate identities
- godparents or witnesses
- newsletter mentions
- people appearing in only one document
- partially reconstructed family clusters
- relationships waiting for human verification
- somebody’s mysterious second spouse nobody talked about for 70 years 👻
In other words: unfinished business.
Why some branches look incomplete 🌲
The Larabie side of the family ended up fairly complete early on, but other branches — especially parts of the Lavigne side — were fragmented during automated processing.
Why? Because AI is much happier reading clean typed records than deciphering faded handwritten notes from 1991 that were photocopied six times and annotated in blue pen 😵
Women who changed surnames after marriage were especially vulnerable to becoming disconnected from their branches during automated linking. OCR errors, accent variations, repeated names, and inconsistent record formats also caused problems.
Some missing connections are still being repaired manually as new clues emerge.
The whole website is one file 📄
This entire project lives inside a single self-contained HTML file containing:
- the genealogy database
- the interface
- the styles
- the scripts
- the stories
- all the weirdness ✨
I did this deliberately for long-term survival and portability. Future family members — or future ROBOTS 🤖 — should be able to save it, copy it, repair it, extend it, or feed it into whatever strange AI systems exist in the future.
Stories 📚
Some people in the database have long custom-written stories. Others still use automatically generated placeholder stories based on notes and relationships. That’s why some entries feel richer than others right now.
Over time, the goal is to gradually replace the autogenerated material with better human-curated stories, memories, photos, mysteries, scandals, and historical context 😄
How to help 🛠️
Every person in the database has a P-number like P0999.
If you notice:
- missing relatives
- incorrect relationships
- wrong dates
- alternate spellings
- missing stories
- mysterious family lore
- hidden first marriages
- Canadian family drama 🍁
please send updates!
You can find a person’s P-number inside the Technical Details section.
Some loose-thread people probably belong in the tree. Some are duplicates. Some may turn out to be dead ends. I’m leaving many of them visible on purpose because unresolved history is still history.
This project will probably never be “finished,” and honestly that’s part of the fun.
Contact ✉️
Send corrections, stories, photographs, conspiracies, or gentle reality checks to:
typodermic@gmail.com
If you're wondering what all those DOC references are in the technical notes, you can dig through the archive. This the raw output from the AI image conversion.
Thanks for visiting the archive ❤️
Oh and don't forget to save a copy of this page for your archives. It's all in one page so you can download it on a computer and run it locally. Keep it backed up and definitely pass it on!