Wow, this actually makes me pretty impressed with 23andme. Yeah, ok, maybe their analysis front-end is borked, but hell, the science seems pretty sound if he was able to get the "right" answer from the original sample.
In my opinion, people who use this service are going to use it because they are genuinely curious about their future health and if it leads to a few false positive, I could live with that and get it checked elsewhere for confirmation.
Libertarian-utopian notions of needless regulatory immurement aside, such requirements exist for reasons amply evident in a study of their origins; in short, without a licensure requirement in such a highly specific and technical realm, incompetents and charlatans of every stripe abound, to the ultimate and severe detriment of all those who would seek the advice of genuine professionals in the field.
All systems have bugs, even the ones which lives depend on. It's cool that the author was able to bring the issue to 23AM's attention, but it's also cool that they were able to correct it.