the current must be assumed to be taking the path of the spring as there can't be guaranteed direct contact between the grounding plate and the button stem. Not without risk of binding. If the stem were to be constricted in it's hole sufficiently to have certain and consistent contact with the plate, it would also be at risk of binding with the tiniest particle present, or with wear and tear. I really wanted that; but couldn't figure a way to accomplish it. I messed with that configuration a lot, and early in the game. When I felt I had to abandon the idea of the button always and without fail being in contact with the plate, and moved on, I built other facets of the design and fabrication process on top of that factor. That is to say, it gets complicated. Between clearances on 3-4 different parts/levels.
Springs are a whole 'nother PITA. Rob's experience with his contact upgrades is a good example. Either you make do with what springs are available. (I still have to modify each spring in three steps/ways before I can use it.) Or you order 10,000 of them, and (apparently) find that your order will be fulfilled in 2-4-6 months . . . . . maybe.
I've been experimenting with odd-lot becu springs that I scored on e-bay. They're stiffer (heavier gauge) on the button press, they're 2" long and cylindrical. They take a lot of modding. Even still, it's a small, limited quantity. I can't provide it as standard.
I'm revisiting the idea of slipping a small (& thin gauge) strip under the spring which would always ensure current path skips the spring. It has to be mechanically weaker than the spring to guarantee a return to the open switch position. And it has to always be in contact with the plate and the button post. Basically, I'm open to ideas on incorporating another part that transfers current directly from stem to plate, and leaves the spring to do only it's mechanical job.
Today, I'm also playing with the idea of a "rocker" switch. Something that is always in contact with the battery, passes through the hole without touching it's perimeter (gasket?), and, when pulled laterally, comes into contact with the plate.