On 8/30/2017 8:14 PM, taf wrote: > Without a paper trail (and that is after all why we are doing this) > you would have to compare all 7 billion genomes on the planet to every > other one, a 50 septillion-fold matrix with a million sites in each > genome to compare. If you have a supercomputer that can do a > million-site comparison in a millionth of a second, it would still > take about 37 million years to complete the comparisons. Once you > start matching sibling genomes and reconstructing hypothetical > parental genomes, each of those would then have to be compared to all > 7 billion. It is ludicrous to suggest such an analysis would take > place, even were the data all to be collected. (and this doesn't even > take the data-loss problem into account) Although the problem would still be daunting, the data would not be entirely random, and there are mathematical and statistical tricks which would make the analysis less extreme than what you suggest. My guess is that if that part of the present thought experiment is not possible with current technology, then it will be before long. On the other hand, getting the necessary samples from every living person on the planet would be orders of magnitude more difficult, and obviously impossible in today's world. Assuming for the sake of argument that the samples were obtained and the analysis were done, and the no-paper-trail genealogy of all of humanity was traced back as far as reasonable confidence would allow, the data loss you have been discussing would cause the tree to fade away to virtually nothing by medieval times.  That much seems clear, but how far back could you get by such methods?  For a given individual, this would depend on numerous factors, and the results would obviously vary.  Based on my experience, I am throwing out what I consider to be reasonable GUESSES, backed up by no evidence whatsoever, as to what would be possible: For individuals with no close living relatives: 0 generations. For individuals with some, but fewer than average, close living relatives: 1-2 generations. For individuals with an average number of close living relatives: 3-4 generations. For individuals with several living siblings and a large number of living cousins on all sides: 5-6 generations. In lucky cases, 7-10 generations in some lines. For populations with significant inbreeding, I would expect these numbers to be smaller. Am I in the right ball park?  Has this "thought experiment" even been studied in any detail? Stewart Baldwin