alzheimer's or the baraka gene for breast cancer and were looking for the five or 10 genes for state educational attainmentq or personality or noncognitive skills and it turns out there's no even a dozen genes, there's thousands if not maybe a million tiny effects all across the gino. if you're trying to discover those teeny effects you need massive sample sizes hence the 1.1 million person genome wide association study. the problem is, once you magnify those small effects in order to detect them, you are magnifying all sorts of biases as well. with the state of literature is low is the 10 to 13 to 14 percent we are predicting we realize now that because part of it is sweeping up a lot of environmental variation that is people, forget about east asia and europe and africa, just in the netherlands there are different frequencies on the north-south axis and east west access. you can probably just as we talked about predicting people self identified race here with 98 percent accuracy based on dna, in the netherlands you can probably locate people in which region of the netherlands they are in pretty darn well