Sorry to barge in on your dismantling of MrV (which I do so enjoy), but I just wanted to say that even though it's science, it does not make the numbers accurate. They are accurate within the full means of scientific testing, but there are plenty of gaps in HIV rate reports. It relies on people who are willfully coming forward to be tested, and those numbers have always been in the minority, especially among heterosexuals.
So, based on your model , we can say that the worst case scenario is that 12 people will be infected, but in the realm of undocumented cases, the numbers are probably much higher.
Remember, the worldwide HIV rate is based on reported numbers, which are gathered from people who get tested; just like the usage rates for condoms come from self-reporting, which people could easily be lying about out of guilt.
This is why statistics are useful as a guideline but not for policy making. You need to rely on ground workers and their anecdotal experiences with these communities to get more accurate information on the probability of higher infection rates.