>>17777925
Vahadou's Multi Source-Target Admixture JS does not detect real ancestry, but rather forces a mathematical model based only on the similarity of the genetic vectors provided, that is, the result depends 100% on the sources you choose and the order/position in which you insert them. If you input any source models, it will always model the highest one as the one with the highest DNA percentage in the target model, even if genetically, geographically and chronologically this does not make sense. The algorithm is simply trying to minimize the Euclidean distance between the target and the reference sources, not to calculate true ancestry. Furthermore, the position of the groups in the list also influences the model, which has already been noticed by several Vahaduo users, showing that the first groups tend to be favored in the modeling if they are loosely compatible.

This means that you can manipulate the results by just changing the order of the provided ancestors. Therefore, any debate that relies solely on this as proof of genetic makeup is invalid: it may be useful as an exploratory or illustrative tool, but it is not definitive evidence of ancestry, especially when we are talking about populations that we do not yet have hundreds of samples from. Using Vahaduo alone as an argument is like trying to see the color of something in the dark using a red light flashlight, the results are biased by the tool itself.