Annealing returns the brass grain structure back to its condition prior to being work hardened. Taking .308 brass and sizing it down to 7mm08 you have to plastically deform the neck i.e. deform it past the point it elastically returns. You can also plastically deform your case neck in firing it (case neck expands to the chamber neck diameter then contracts a little) and then sizing it back to be reloaded again (neck diameter is reduced to allow seating of the projectile with some tension). This plastic deformation work hardens the material, too much work hardening and the material becomes brittle and will easily break. Think of it like a paper clip that you bend a few times until it breaks... this is work hardening.
Plastic deformation deforms the material grain structure. The brass grain structure has to stretch and elongate to allow for the deformation, too much of this and the grain structure will crack. Heating the material up above its annealing temperature (but below its melting temperature) will allow the grain structure to re-crystallize (return to its pre-deformed structure).
Air cooled vs water cooled. Water cooling rapidly cools the brass compared to air cooling. The more rapid the cooling the harder the material ends up... and the less ductility it has.
Personally I resize .308 brass down to .260 and I have not found the need to anneal it yet. That said the chamber in my .260 is tighter than the SAAMI spec and I have to neck turn my re-sized brass. As a result I run very close tolerances between my chamber neck diameter and my loaded brass neck diameter. In running such close tolerances the neck on my brass does not go through much deformation either when it is fired or when I neck size it to be reloaded. Some of my brass has been reloaded nearly 10 times now and I have yet to throw one out.
Bookmarks