It is not hard to take a listing of the single nucleotide polymorphisms (SNPs) in a persons DNA and compare them against a database of genetic sequences from various parts of the world to tell where they came from. The same mutation occurring in parallel in two different populations and persisting is rare enough that you can use SNP data to construct a phylogenetic tree of the human population (here is one I googled up[2]). We can tell interesting things from human population genetics, for example that Amerindians probably came from Asia, since the East Asians are there closest genetic cousins.
You should have a mixed-race friend send their spit into 23andMe and see what happens! When races mix, you have genetic admixture. It remains trivial to see which portions of a persons DNA comes from which part of the world because those portions contain SNPs which are only found in certain human subpopulations.
If you are the child of an African father and a European mother (like Barack Obama), you get half your chromosomes from your father and half from your mother. A few bits from each chromosome swap over, but it is not hard to unravel which bits came from Africa and which from Europe.
The government's big racial categories are somewhat inaccurate (especially hispanic, since some hispanics are European and some are Amerindian). But distinct human genetic populations are very real. Razib Khan's blog[3] is great if you want to move beyond the sound bytes and get into the real science of human population genetics. Sadly, liberal views on race tend to be extremely unscientific.
[1] http://www.forbes.com/sites/jonentine/2012/08/12/the-dna-oly...
No, it's the composition of the species' genomes. If they were created in a lab yesterday the situation would be the same. Unless you believe in some kind of supernatural effect, the way homeopaths believe that water has "memory".
"we see no sickle cell disease in Europeans as we do in Africans"
There is sickle cell disease in southern Italy, and in many other parts of the world outside of Africa. It cuts across "racial" divisions, and follows the historical distribution of malaria.
You avoided the simple question about 23andMe. You claimed they would have "no problem" categorizing people as "white" or "black". So how would they categorize the child of a "white" and "black" parent?
"23andMe would have no problem telling them apart"
If someone you define as "black" marries someone you define as "white" and their child sends a sample to 23andMe, will the report say that the child is white or black? Remember, you claim they will have "no problem".
Andrew Sullivan, from the link you provided: '“race” is a social construct when we define it as “white”, “black,” “Asian” or, even more ludicrously, “Hispanic.”'