Tin Chest
A family history that shows its working.
Most family trees are confident and wrong. This one records what every claim rests on, how sure it is, and what contradicts it.
The problem
A tree on a big genealogy site is a pile of assertions with nothing attached. One person guesses, three others copy the guess, and the guess becomes a fact by repetition. Nothing in the format can tell you which parts are evidence and which are somebody's Tuesday afternoon.
The method
Every statement in the archive is a claim. A claim names its source, carries a tier and a confidence, and shows its reasoning. Claims that contradict each other both stay. Identity is provisional until something makes it certain, and the archive says so in writing.
- tier
- record | inference | hearsay
- confidence
- high | med | low
- reasoning
- written down, at the time
A worked example
Five points, not the surname.
In 1841 there were 147 Halliwells in a Wigan of 25,524 people. One person in 174. A surname match there is worth nothing, so this household was identified on five independent points from five different sources: the occupation, a wife's age, a child's age, a street against a baptism address, and a grandmother whose stated age was wrong in exactly the way the 1841 rules required.
clm-c41-03 · present-at "the same household, aged 65"
when 1841-06-06 · tier record · conf high
src src-1841-census-wigan-lan-opc
The 1841 census rounded every age of fifteen and over down to the nearest five. Alice was 79 in March 1851, so she was 69 in June 1841, which the enumerator was required to write as 65.
The archive holds more than 7,700 claims like this one.
What is coming
Articles on doing this properly: reading a census transcript, building a claim grammar, keeping contradictions instead of resolving them, and putting research agents to work without letting them invent ancestors. Then the tool itself.
Get the articles
No schedule, no newsletter chatter. New writing only.
We only email you new writing. See Privacy.