In the most recent discussion, the terms to distinguish based on ethnicity are "whites" and "people of color". Basically, Eastern-european descent and everybody else. I suppose, when dealing with diversity there has to be a description (hopefully relatively accurate) of each of the discriminated groups, but I don't really like the "people of color" classification, because everyone, except people with albinism, and other melanin-disorders, have melanin, and hence, color, in their melanocytes (skin cells). It seems just as divisive, and just as inaccurate, as saying "black", "yellow", and "red".
One solution would be to have a scale measuring the level of pigmentation, based on the melanin content in ones skin cells, say an average measurement of three locations on the body, using photographic technology. A problem with this would be that people tan, and the range for various ethnic groups overlaps. However, it would be a step closer in the right direction. Plus, we could have more than two classes - it could be measured on a continuum. Especially because there is evidence that even among different ethnic groups there is discrimination based on the relative darkness of skin. A measurement based on range of pigmentation would help discern this.
Another idea, with the advent of cheap, efficient, sequencing technology, is to run a very basic genotype of each person, say, when you go to get your photo id from the government (driver's license or whatnot), looking for markers indicating which ancestral populations lie in your genome, and each person could report the number of markers they have relative to each of the studied ancestral populations. This would be more accurate, but perhaps not always as effective for researchers, because some discrimination is based on what people see, and heritage is sometimes hidden in our genomes.