I've always soaked the entire hinge with thin CA. Never had a problem and although in theory it sounds plausible I think if any sort of experiment with valid imperical data were conducted you would find it next to impossible to NOT get CA on the bending portion of the hinge. Keep in mind, the hinge acts as a WIC and will draw the CA in all directions. I can not imagine a way to glue the hinge from the and not have it wic in all directions from the area where CA is applied to (directly adjacent to hingline where it touches balsa. If you're not putting enough CA to wic 1mm backwards you're definately not putting enough to wic 1/2" inwards into the wood. You HAVE to glue them from center outwards, it's by design.