First, a few facts:
- On average, the genomes of two people vary by only 0.1 percent. That is, you are genetically 99.9 percent the same as everyone else.
- About 85 percent of genetic variation is within populations, that is, each large population (Japan, say) contains about 85 percent of human genetic diversity. So only 15 percent of variation is between populations, and can be used to discriminate between them.
- The human genome contains about 3.2 billion base pairs, and 15 percent of 0.1 percent of 3.2 billion is still around 500,000, which is quite a bit of variation to work with. But you should never forget that the differences we are talking about are tiny compared to the similarities.
- Differences between different parts of Europe (Ireland vs. Norway vs. Poland) are even smaller, since all Europeans share some genetic characteristics.
- All information on population genetics is statistical, so all conclusions should be expressed as probabilities, not certainties.
- The quality of the result depends on the quality and quantity of the data. Right now the databases are weak everywhere outside Europe, especially in Africa.
- Much of the data used by private companies is reported by their own customers, that is, they compare their results to where their customers say they come from. So if a lot of people are wrong about their genealogies, that would corrupt the results.
I write about population genetics and human history here all the time, because it is very exciting new science. But new science is iffy science, and I try to be clear that all conclusions in this discipline are tentative and subject to correction as our knowledge improves. So go ahead and order a genetic origin test if you want, but don't put too much faith in it.