Commercial genetic ancestry tests can assign percentages to regions with impressive precision. Those percentages depend on reference databases, statistical models, and changing population samples, so they are estimates rather than fixed maps of identity. Isolating one detail would make the passage seem chiefly about the fact that ancestry percentages depend on reference databases and statistical models, but the author uses that detail to defend a wider claim: genetic ancestry results can offer clues about lineage but should be read as model-dependent estimates rather than definitive identities.