For a good look at the maths, see this blog post: http://www.goodmath.org/blog/2015/02/05/quick-vaccine-math/
Infections spread exponentially, with each infected person infecting X other people (I'm oversimplifying here of course). The key is to make infection of a random person sufficiently unlikely that X < 1, so the infection dies out rapidly instead of becoming epidemic.
No. Many vaccines are very effective, but not 100% effective. In addition, while vaccines are generally safe, there are people people that are not "anti-vaxxers" (or their children) who are not vaccinated, including those two young to receive a particular vaccine or those who do to particular health condition cannot safely receive a particular vaccine. General vaccination not only reduces the probability (but not to zero) of those vaccinated getting sick if exposed to the disease, by doing so it reduces the probability of them exposing others to the disease.
If vaccination is not general to all those who can safely receive it, its effectiveness in preventing disease both in those who are vaccinated and in those who are not vaccinated is reduced, because both the vaccinated and the non-vaccinated are more likely to be exposed to the disease than they would be if vaccination was general.
>So if the rest of us are immunized the only people getting sick are the anti-vaxxers.
You are leaving out the infants that are not yet old enough to immunized as well as those with immunodeficiency that cannot be vaccinated.
So if the rest of us are immunized the only people getting sick are the anti-vaxxers. For this allegory of the Car Brakes and Social Responsibility to truly be a 1-to-1 comparison of social risk with regards to pro/anti immunization he'd have to basically present the premise that everyone else who has car brakes is protected from the less informed anti-brakers.