Genealogy is equal parts story and systems engineering: you can’t write compelling narratives if the underlying data is messy.
So I periodically export my database to GEDCOM and do a quick “health check.” Here’s what the latest export reveals.
Tree size (current export)#
This GEDCOM currently contains:
- Individuals: 27,840
- Families: 9,307
- Sources: 1,040
- Media objects: 1,555
That’s a lot of people—and exactly why consistency matters.
Top surname clusters#
Counting surnames as recorded in NAME entries (which can include variant spellings and alternate names), the largest clusters include:
- Wheeler — 1,936
- Woodbury — 997
- Worcester — 350
- Tuttle — 270
- Baker — 256
- Rice — 250
- Tolles — 244
- Smith — 227
- Beasecker — 209
- Dodge — 199
This is useful for prioritizing what gets attention first—both for cleanup and for weekly story releases.
The “quiet” problem: surname variants#
GEDCOM exports make variant problems painfully obvious. Here are two clusters where the same line is currently split across multiple forms.
McGinty / Ginty variants#
- McGinty — 89
- Mc Ginty — 5
- Mcginty — 4
- Ginty — 21
Mackenzie variants#
- Mackenzie — 106
- MacKenzie — 7
These variants matter because they affect:
- tag pages (and what “counts” as the same family line)
- search quality
- duplicate detection
- exports and reports
If you’re researching these lines, try multiple spellings in Search.
Place hubs: where the paper trail clusters#
Place fields are inherently messy (“England” vs “Town, County, State”), but even with inconsistency you can still see strong hubs.
This export shows recurring concentration in:
- Massachusetts (especially Essex + Middlesex County towns)
- Midwest cluster: Indiana / Ohio / Illinois (Chicago)
- England (often recorded generically as “England”)
- A notable cluster in Mazowieckie, Poland (appears frequently in place fields)
This is why the site has dedicated Places and Surnames navigation now:
- Surnames: /surnames/
- Places: /places/
- Browse hub: /browse/
Data quality checks (because genealogy runs on receipts)#
A couple quick “sanity checks” surfaced:
- Birth-after-death flags: 24 records where the first captured birth-like year appears later than the first death-like year
(usually a swapped date, wrong attachment, or merge artifact) - Famous/historical duplicates: multiple repeated identities (common after merging/importing)
- Place formatting inconsistency: the same location appearing in multiple formats
None of this is unusual. It’s just normal maintenance once a tree gets large.
What I’m doing next#
- Normalize the big surname variants (McGinty/Ginty; Mackenzie/MacKenzie; etc.).
- Standardize place formats for the biggest hubs (MA towns/counties; Chicago IL; key Ireland/Poland locations).
- Use the cleaned dataset to drive weekly story releases by branch.
If you see a surname split or place variant that should be unified, send me a note.
Contact: /contact/