The FDA's theory here, though, is that since 23andme goes beyond just saying "You have the G:A variant of Rs601338" to "You are probably as susceptible to norovirus infection as a typical person" they're not just a medical laboratory but also a medical device manufacturer, and therefore come under the FDA's purview with much stricter tests for what they can market and what sort of tests their product undergoes.
To me, that sounds pretty ridiculous since you can easily just go on SNPedia to look up the associated information yourself[2].
[1]http://www.cms.gov/Regulations-and-Guidance/Legislation/CLIA... [2]http://snpedia.com/index.php/Rs601338
Would you want a physician picking your drug regimen based off of what Wikipedia says on some particular day? Would you like your implanted defibrillator to change its algorithm based on something Wikipedia says?
If not, then you've discovered the crux of the issue. There's a bright line between "data that someone put online yesterday" and "rigorously tested and proven effective". That's the whole point of FDA regulation on what gets to be marketed as a medical device, or what drugs can be prescribed for an indication.