There an article in this month's Nature examining the statistical evidence for universal common descent. This is the first time someone has taken the massive amounts of genetic data and applied a Bayesian analysis to determine whether the existence of a universal common ancestor is the best model. Most of what we generally think of as evidence for evolution and shared ancestry is evidence for shared ancestry of large collections, such as mammals or birds, or for smaller groups. Some of the evidence is for common ancestry for a phylum. There is prior evidence for their shared ancestry based on primitive fossils and on the shared genetic code and extreme similarity of genomes across very different species. This is the first paper to make that last argument mathematically rigorous. When taken in this fashion, the paper more or less concludes that a Bayesian analysis using just the genetic and phylogenetic known data puts the universal common ancestor model as overwhelmingly more likely than other models. (The article is behind a paywall so until I get back to the university tomorrow I won't be able to comment on this in any substantial detail but this looks pretty cool and a good example how careful Bayesianism can help make something more precise).