Skip to main content
The Reading Room

The pipeline

What we did with your file

Before you can ask a single question, every record in your file passes through a pipeline that cleans it up, protects the people still living, and writes a short, plain summary of everyone in it. Nothing below was written by a model — every number and every example is exactly what the pipeline actually did to your file.

royal92.ged0MB · 30,682 linesGEDCOM unknown · utf-8every line read

At a glance

3,010

People

1,422

Families

5,592

Life events

0

Sources cited

670

Distinct places

98%

In the main tree

4 smaller, unconnected groups besides.

How we protected the living

706 of the 3,010 people in your file may still be alive. We hold back their details — birth dates, addresses, family links — from search results and summaries by default. When we could not tell either way from the record, we treated the person as living rather than guess: a person wrongly shown as living is an inconvenience; a living person’s details published by mistake cannot be undone.

They are still read into the tree and placed on the meaning map, because a family missing its most recent generations is not your family. Being in the index and being quoted back at you are two different decisions, and only the first one is on.

What we cleaned up

Family history software writes dates and places in dozens of inconsistent ways. We made sense of 4,016 of the 4,018 dated events in your file (from 2,111 different ways of writing a date). Your file also referred to places 715 different ways; we resolved those down to 670 actual places, including 3 pairs like "Little Rock, Arkansas" and "Little Rock, Pulaski, Arkansas" that were the same town recorded at two levels of detail.

A few places, as your file wrote them

London,EnglandLondon, England
,,London,EnglandLondon, England
London,,,EnglandLondon, England
London,,EnglandLondon, England

A few dates, as your file wrote them

12 MAR 1637/163812 March 1637day
15 SEP 1396/139715 September 1396day
ABT 14 AUG 1479about 14 August 1479day
BEF 16 FEB 1337before 16 February 1337day

What we flagged — and never changed

We never delete or silently correct anything in your file. Where a record contradicts itself — a death before a birth, a parent younger than their child — we leave it exactly as written and attach a note, so you can look into it yourself.

We also found 5 pairs of records that might describe the same person — never merged, only suggested.

Henry_(1) Tudor and Henry_(2) Tudor

84% confidence — same parents and same name, same surname, same given name, born 2 years apart, same birthplace

Why this makes it cheap to ask questions

Your original file is about 117K tokens — far more than any conversation could hold. Because we did the cleaning up front, a summary of one person now costs about 36 tokens, and answering “how are these two people related” costs less than a hundred — computed directly from the family structure, never guessed at.

117K

Raw file, in tokens

36

One person's summary

3,257×

Smaller, per person

See your treeSearch for someone