I don't have the background to have an independent opinion, but I have used almost all of the admixture calculators and looked at the underlying studies briefly. One thing that is often absent in our discussions is that the studies are not about genealogy but about population genetics, and are concerned with the development and movement of large human populations in the context of biological adaptation. They have been adopted for use as a supplement to other genealogy methods, but were not originally designed for that purpose.
I consider the ethnicity analysis to be one of many potentially useful tools. I've complemented my traditional research with YDNA, MtDNA, autosomal DNA and admixture tests and found, taken together, that they provide a fairly coherent picture of my deep ancestry and some interesting clues about my family in genealogical time.
I recently received my results from the Geno 2.0 test and am looking forward to seeing Spencer Wells' forthcoming paper.