AO brigade: 0.80*(100)*(0) + 0.20*(100)(1) = 20 casualties.
BPV unit: 0.80*(100)*(0) + 0.20(100)(.25) = 5 casualties.
See the problem that most AO proponents have is that they assign 100% *effectiveness* to the procedure of abstinence, but they forget to assign a probability that a given person will remain abstinent. That is, there are really two different issues here, by analogy with bullet proof vests. The first is the effectiveness of the vest in stopping bullets. The second is the likelihood a person will be shot. We can't assume that there is a zero percent chance of being shot, even if we do our best to avoid it.
Usually the AO proponent will start attacking the analogy. No analogy is perfect, because if it were identical to the situation being compared it would no longer be an analogy. It would just be a restatement of the original situation. They want to nitpick this or that feature of the analogy, because they can see that it is clearly fatal to their position. Perhaps the best approach to all this quibbling is to generate new scenarios.
For example,consider this one. Instead of sticking with bullet proof vests, consider vaccinating for an infectious disease. The AO crowd would be arguing, analogously, against vaccinations, just as religious people have opposed vaccination in real life in the past, because vaccinations are not perfect. Instead they would be taking the view that the best way to prevent a disease is to abstain from exposure to the germs that cause it. So long as one can do that, one will never catch the disease. Alternately, pro-vaccination folks are arguing that one needs protection for the times that one cannot reasonably expect to avoid these germs.
The AO proponents naturally feel that sex, unlike germs or bullets, is something that is much more controllable, and that is something I am prepared to grant, in many cases. However, they apparently are unaware of how many date and acquaintance rapes take place. There is, of course, rape by strangers as well, but those are not as common. There is also the possibility of drugs impairing one's ability to abstain. There is even just succumbing to temptation, which happens even with saints. Because of all those factors, one MUST assign a non-zero probability that a person's attempt to abstain from sex will fail. When AO's say that abstinence works 100% of the time, they are wrongly assuming that every person has 100% control over sex in all circumstances. Even if we assume that, under normal circumstances people had 90% control, when we are talking about a policy for hundreds of millions or billions of people, that 10% failure amounts to a lot of people.