HIV has a few limits, firstly (I think) it only infects one type of cell (CD4 immune cells - white blood cells), secondly, it doesn't spread as easy as influenza, you need to exchange some bodily fluids with someone, preferably with an abundant supply of CD4 cells to infect. Since it generally requires sex to get HIV, if you knock out 91% of strains the other strains wont rush in to fill the vacuum.
The resistant strains already present will almost certainly increases as a proportion of the HIV population, its just natural selection. Use of condoms can prevent people from getting the resistant strains. Interestingly, male circumcision is also an excellent way to reduce chances of HIV infection because there will be less CD4 cells in your penis.
Plus, one of the big problems thus far have been we had NO IDEA how to attack the HIV bodies. Perhaps developing an understanding of this antibody will allow us to gain new understanding, and perhaps in good 'ol engineering tradition we will be able to change this, that, recompile, and cover a few more %. I know it's not that easy, but if you've ever developed code or hardware, having pre-existing, working models makes everything miles and miles easier.
That is inaccurate. Current antiviral agents remove most (as in over 90%) of the HIV viral particles from the blood of a patient. But then after weeks or months the viral load in blood is just as high. And any given patient has numerous strains within. Furthermore, most viral particles are not active: the virus saturates the immune system with defective particles (in a way it screens itself under noise). What's worse, HIV has long-term repositories like nervous system cells, from which particles are emitted at low rate. Killing your brain cells just to get rid of HIV defeats the purpose.
What this novel antibody will do is to remove 91% of the strains, meaning the remaining 9% will flourish enormously . RNA viruses mutate very fast and broadly (always being one step before crash-error).
What a good antibody can do is cut short the epidemics side of HIV: if a patient's HIV load is so low that it can't effectively and consistently pass on its infection to another person, the infection dies with the patient. This leads to a negative feedback loop that stops the epidemic. It's the same concept as vaccines: ensure that a sufficiently large proportion of the population is immune, so infections cannot spread when they inevitably appear in isolated individuals.
The problem with HIV is that every obvious way of attacking it seems to have some kind of drawback that's less than obvious. You can sit around and think up all kinds of hypothetical scenarios for attacking it and find that people have already tried that. And failed.
http://news.ycombinator.com/item?id=1496772
EDIT: sorry, I worded this terribly. What I meant was there are many people who would not catch HIV again, and thus would benefit from a cure to only 91% of HIV. The people who will continue being reckless and catch it again will, of course, not benefit from an antibody protecting them from only 91% of HIV, and that was what I was talking about when I referenced druggies. I'd be more than happy to give hardcore druggies a cure.
The whole point was not that some people deserve it and some don't, but that some will gain huge benefits even if it's only 91%, and that is why even though it's not 100% coverage it's still a plus. (of course, not everyone will, because some people will go and catch the other 9% right away, so for them it's not really a benefit, but SOME WILL BENEFIT)
EDIT2: to those of you still downvoting, are you trying to say you think every single person on the planet with AIDS would not catch it again if they were cured, but still vulnerable to some strains? That seems kind of unlikely.
edit: feel free to re-word it if you like, as long as you leave a note that you edited it.
That reminds me a lot of Good AIDS vs Bad AIDS from the British documentary satire Brass Eye (http://www.youtube.com/watch?v=kFNs2mOkKzc)