I'm sorry but that summary, the same as in the document you link to, does not actually say that HIV was lower in the circumcised men than the non-circumcised, only that it was lower after the circumcision campaign. This is the fallacy of "post hoc ergo propter hoc" - after this, therefore because of this. We may certainly hope that circumcision was not carried out without a concurrent campaign of education of both the circumcised and non-circumcised men, about the use of condoms etc. If it was, we have no way of knowing that it was circumcision that made the difference. I should like to see Auvert's calculation that it was. As I mentioned, Weiss, Gray, Auvert, Hankins, Dickson et al. are all members of a remarkably small group in the world scene dedicated to promoting circumcision. (There are papers carrying all their names, along with Bailey, Halperin, Morris, Klausner and a few others we see again and again on this topic.) Because of the enormous amount of cultural baggage it carries, circumcision is not like other surgical interventions, and people's passion for it is not always rational. It is more than usually subject to the human biases that can beset any branch of science, and therefore need needs more than the usual amount of correction for these biases. This it has not had.