Actually, I don't remember any such thing. Perhaps you'd best diagram the relevant sentence in the abstract, which is all we have of the paper.
As for Morton's demon, if you're claiming that both of us are afflicted, fine. You will need to make far more precise arguments than just making broad claims. I will attempt to do the same.
Follow the links I have given. The paper is free for viewing.
Here is from the paper.
In principle, the following gene conversion scenario could have resulted in the removal of the provirus in the human lineage. First, the preintegration locus underwent a duplication event in the common ancestor of Homo, Pan, and Gorilla. Second, the provirus formed in one of the two copies of the locus by viral infection of the common ancestor. Next, the Gorilla lineage diverged from the Pan-Homo common ancestor. Then, the Pan and Homo lineages diverged. Afterwards, a recombination event reversed the original locus duplication, restoring a single copy of the locus without the provirus in the Homo lineage. However, the PCR and sequencing assays uniformly failed to detect any evidence for the presence of such a duplicated locus in Gorilla or Pan. Therefore, in addition to the removal of the provirus specifically in the Homo lineage, the scenario also requires recombination events in the Pan and Gorilla lineages to eliminate the provirus-free copy of the locus. Since the Gorilla lineage diverged before the Pan and Homo lineages separated, this means that independent recombination events would have had to occur in both the Pan and Gorilla lineages. While this scenario can never be formally excluded, there is a more parsimonious alternative that involves three fewer recombination events.
In the above they try to fit the outcome to the tree but it is not believable.
The alternative is an allelic segregation model (Figure 2d) in which the provirus formed in the most recent common ancestor of Homo, Pan, and Gorilla just before the three lineages separated. The provirus allele was fixed in the Gorilla lineage. Both alleles were then maintained in the Pan-Homo common ancestor until the individual lineages diverged. The provirus allele was fixed in the Pan lineage, while the preintegration site allele was fixed in the Homo lineage.
That is the fitting of the data to the desired tree. And my tables show how any result of the tests can be fit into the above described tree.