Your Ancestry Result Changed and Nobody Edited Your DNA
Genetic ancestry percentages shift when companies update their reference panels, and that is not a bug. Here is what an ancestry estimate actually measures, why 'genetic ancestry' and 'ethnicity' are different things, and what the number can honestly tell you.
Someone takes an ancestry test. A year later the company updates its method, and the customer is now 12% less Scandinavian than they were at breakfast. Nobody touched their genome.
This is the most common complaint about consumer ancestry testing, and it is worth understanding, because the answer explains what these tests are actually doing.
What is being measured
An ancestry estimate is a similarity measurement against a chosen set of reference populations. Your genotype is compared with panels of people whose ancestry is treated as known, and the output is the mixture of those panels that best accounts for your genome.
Every part of that sentence is doing work. Chosen set. Treated as known. Best accounts for.
Change which populations are in the panel — add more Finns, split "Southern Europe" into three groups, include newly sequenced ancient genomes — and the percentages move. Your DNA is constant. The yardstick is not.
Why "ancestry" and "ethnicity" are not the same
Genetic ancestry describes the populations your DNA segments most resemble. Ethnicity is a social and cultural identity, involving language, religion, self-understanding and history.
They correlate, sometimes strongly. They are not interchangeable, and a percentage cannot adjudicate an identity.
There is a further wrinkle that surprises people: you do not inherit your genealogical family tree evenly. Go back ten generations and you have 1,024 ancestors, but only a few hundred contributed any detectable DNA to you. Recombination is a lottery. A great-great-great-grandparent can vanish from your genome entirely while remaining, unambiguously, your ancestor.
Ancient DNA rewrote the framework
For most of the twentieth century, human population history was reconstructed from present-day variation, working backwards. That approach carried a hidden assumption: that today's populations sit roughly where their ancestors sat.
Ancient DNA broke that assumption. Populations turned out to have been replaced, merged and displaced repeatedly, meaning modern groups are frequently poor proxies for the ancient ones they geographically overlap (Pickrell & Reich, Trends in Genetics, 2014, PMID: 25168683).
At the same time, genetic variation across Europe does track geography with real precision — a genetic map of European samples reproduces the physical map of Europe remarkably well (Novembre et al., Nature, 2008, PMID: 18758442). Both things are true. Geography leaves a signal; history keeps moving the people.
How to read your own result
The Ancestry app, built on Joe Pickrell's algorithm, compares your genotype against reference populations and reports your estimated proportions - where your genome would have been placed within that reference set, not a fact about who you are.
Read it as: this is which reference groups my genome most resembles, given this panel and this method. Do not read it as a statement about nationality, belonging, or a percentage of who you are.
And when a future update shifts your numbers by a few points, that is the method improving. It was always a measurement with error bars — most interfaces just decline to draw them.