Error in the age of personal genomics

Over at Genetic FutureDr. Daniel MacArthur points out some errors in deCODE's interpretation services. Dr. MacArthur presumably knows his maternity, though if the X chromosome results were correct one would guess that Dr. MacArthur is actually adopted and that his mother might be a Lumbee Indian. But it makes me wonder how confused people are going to be due to problems with false results. In particular, as these technologies become very cheap many families with make recourse to them. Sometimes this will highlight "extrapair paternity events," but sometimes there will be errors and siblings may face a period of uncertainty in relation to possibly discordant results. The likelihood of a false result creating an unexpected situation is conditional on various probabilities, the error rate of the results, and the likelihood of an extrapair paternity event (which varies from demographic to demographic and family to family). I guess we'll have more data in the near future....