There's no way that study is saying there's a 17% chance of reinfection 5 months after recovering from the first infection - the number of people who have been infected twice (by different strains) is infinitesimally small, only a couple dozen known cases out of tens of millions of infections. They must be measuring antibody levels, but that's the wrong way to look at it - B and T memory cells are what provide lasting protection, you don't need to be making a bunch of antibodies constantly to something you're not being exposed to, it's a waste of your body's resources.