Another point: if you carefully study the tables in the "accuracy and precision" section of the ABD website, you will see that while 2.5 is superior to 2.0 in accuracy of the IndoEuropean/Native American/East Asian distinctions, 2.0 was superior to 2.5 in the accuracy of African/nonAfrican distinctions. In this regard, the decision to discontinue 2.0 is unfortunate.
Validation: there is in America much admixture. There is a lot of opportunity to find predominantly IE people who are known to really have Chinese grandmothers and Native American greatgrandfathers, etc., and observe if the test gives reasonable MLEs