I would argue that today it is already proven that we can guess with better than random accuracy. For example, a significant percentage of people with Autism have an identifiable genetic cause, and our ability to identify genetic causes for Autism is only going to increase over time. Studies have already shown that Autistic people are statistically significantly more likely to be atheists than non-autistic people. It seems to follow that if we identify DNA with genes indicating Autism, that we can guess it belongs to an Atheist, and that we will achieve better accuracy by taking the genetic information into account than we would by assuming the probability of atheism is that of the general population. A. Autism can sometimes be identified by genetic analysis B. Autistic people are statistically far more likely to be atheists than non-autistic people therefor C. In at least some cases, genetic analysis can be used to statistically determine that the probability of Atheism is greater than it is in the general population If points A and B are true, and all of the scientific literature I have read seems to indicate that they are, then point C seems to logically follow.