The program we usually use is called STRUCTURE. It's likewise the program where nearly all of the data from your link came from. STRUCTURE is hardly equipped to perform objective scientific analysis--it is, like many things in social science, a program which draws upon some scientific data to infuse it with social hypotheses and answer a specifically formulated, purposeful question. We can't use STRUCTURE to figure out true things about reality, the way we can use astronomical analysis packages to derive conclusions based on the data that a space probe collects. STRUCTURE requires a human element--the data, on its own, speaks for itself, as far as it can. We need to tell STRUCTURE what to look for, and in doing so, we tell it what is important to us, personally.

Take racial clusters, for instance. Those uneducated in genetic sciences see your link and assume that racial clusters are real, that they are true, and located in the data. They mistakenly believe that STRUCTURE draws this truth out of the data for all to see. However, STRUCTURE has no way to determine the correct number of racial clusters. We actually have to tell it how many we want to see. If you want to believe that there are 7 racial clusters, you can tell STRUCTURE to look for 7. It will find 7--after all, that's what you told it to do. If you ask it to find 12, it will find 12. The simplest operation is to tell STRUCTURE to find 1 cluster, and this is presently the most widely accepted number of racial clusters that exist. Sometimes the data looks more visually appealing to us, looks like it 'ought' to be 7, or 12, or 23. But we can always use a sharper magnifying glass, or take a step back, and see that, in terms of genetic science, there are only as many 'races' as we choose to see. Usually, we choose a certain number because we have a specific question, often epidemielogical, that we want to answer. The 'correct' number of races in each instance is whatever helps us answer that particular question.

I should point out that I am providing you the courtesy of pretending that 'race' is a legitimate taxonomic category. It is not. 'Race' typically refers to subspecies, which is likewise not a defined classification. There are hundreds of studies that use hundreds of different definitions of what qualifies as a subspecies. Historically, the definition has only had glimmers of consistency across specific areas of research, for specific species and genera. For example, wolf researchers tend to use 'subspecies' in the same way, because they cite other wolf researchers who used it that way. It is a totally different story for drosophila researchers. So, again, before we have started, your premise is non-scientific. But I don't even need to win on that point--this is my job, I could give you free points all day and still win.

You should be aware: your link is very misleading, and has hoodwinked you. These 'ethnically dependent' SNPs? My, that sounds impressive! Damning, even! How could I possibly argue with 'ethnically dependent' SNPs? Easy.

These SNPs are specifically derived from unexpressed remnants of viral DNA (retrotransposons) that mutate very rapidly. Because they do not have an effect on our phenotype, these mutations are not 'pruned' by natural selection. They are then able to proliferate and diversify, and allow us to compare samples of aDNA (ancient DNA) to modern samples, and match up who is related to what groups based on the pattern of mutations.

Do you see an obvious problem? Humans are so incredibly closely related, so lacking in genetic diversity compared to most animals, that we have to go far out of our way to be able to detect differences at all. We actually need to look at genes that don't do anything. We need to look at genes that have no effects on us, because if we try to find genetic diversity elsewhere, we come across too many stumbling blocks. Your 'ethnically dependent' SNPs, the keystone upon which your link depends, are quite precisely the LEAST MEANINGFUL genes in the human genome. That isn't a coincidence. That's the only way we can reliably distinguish our personally preferred number of races--by looking at genes that don't do anything. You are using genes that don't do anything and saying that they enable us to distinguish race. If race is so self-evident, why don't you look at genes that DO things? Because you can't. The analyses will be inconsistent. You would have to pick and choose genes that make your point, and ignore that vastly higher number of genes that don't. Or, you could perform a genome-wide analysis, which will put you in exactly the same position--the differences will be so small relative to the whole of the genome that, by definition, they will fail to meet the standard of statistical significance. I've been doing this for a long time.

I'm not naive on this science. There are certainly genes that have an effect on IQ, testosterone levels, impulsivity, etc. These genes are predictably distributed in various popilations. We all know where they are prominent, and where they aren't. But you aren't talking about that. You are trying to shoehorn in Biblical 'kinds' into modern science, under the same guise of 'race' that the racialists of the 19th and 20th century used. Your fundamental hypothesis is Biblical, not scientific. And, the anthropologist in me sees your conflict with the Jews for precisely what it is--family squabbles. Jealousy of the more 'successful' big brother, who isn't letting you in his clubhouse.

To be clear, you led with your BEST argument. I didn't characterize it as such, you did. I demonstrated why that argument is, scientifically, nonsense. It is one I have heard many times, and proliferates on boards like this. It is actually a running joke in the genetics community.

Pleaze recognize what has happened here. Your BEST argument (your characterization) was bunk. Total drivel. Useless. 'Pseudoscience' is too decent a label for it. This is why I was hoping you would invoke machine learning as a means of possibly determining an objective k (cluster) value. But we didn't even get that far.