On many/most props, the adapter ring goes in just on one side (underside, facing motor on a multirotor) and doesnt go through all the way. That underside usually is also nicely flat while the other side tends to be tapered; it makes more sense to me to use the taper of the washer to help center the prop from above as well, and you may even get a larger friction surface depending on how compatible the taper slopes are.
As for tearing it apart that way; in theory you would be right, but from the last 100 or so props I killed, I dont recall I ever managed to destroy a propellor hub, so I wouldnt worry too much about that.
Anyway, you may be right, but I guess either way will work.
As far as the ring on one side, it generally does go on the machine flattened side of the prop, so it should be held square with whichever side it pressed against. Compression should then ensure centering is maintained by forcing everything to square on both ends of the shaft. As far as which side of the adapter gets the tapered side of the prop . . . that will depend if it's being mounted as a pusher or a tractor. Even pusher props have the machine flattened side on the dirty-air side of the prop . . . where the washer is supposed to be.
If the tapered slopes are machined to match angle, then I'd agree -- they are designed to mate -- but the propeller manufacturer is seldom the adaptor manufacturer so those angles are not coordinated -- a standard "countersink" angle does not exist, so far as I'm aware (I've seen plenty of dimensional diagrams of prop and have yet to see the non-machined side specified in any way). As far as getting better area, I'd rather have the washer have more grab farther out radially than near the center -- 3mm of cross-sectional contact length near the shaft is far less area than 3mm of cross-sectional contact near the outer edge of the hub. In a perfect world with props matched to adaptors, the washers would be self centered with clean contact from the inner edge of the washer to the outer. Clearly this isn't the case.
This convention for "low torque on the fastener vs high torque on fastened" using a cone-shaped washer does extend outside the toy airplane prop adapter realm -- I've seen this applied to machines long before I saw them on props. There are engineering applications where the surfaces can be reversed in good practice, but generally only when the surfaces are machined to precisely match, and the cone is not flattened at one end.
I do agree that the "tearing it apart" is a minor effect -- the taper on these is below 10 degrees so the radial force will be less than 20% of the axial force -- but I've often wondered about many of the larger multirotors shedding props are more from defective prop castings or some encouragement from the prop mounting. Minor effects will show up as the root cause from time to time, but few will believe it's the root simply because it happens so infrequently. I've got no evidence, so even I take it with a grain of salt.
In the end, I agree -- it *WILL* work either way -- however one orientation will require a greater torque to properly fasten. As I have mentioned in other threads there is an issue with people under-torqueing the adapter for fear of shattering the prop (if it does, they need to buy better props!). So long as it's tight enough the bell and prop can't be slipped relative to each other, how you got there is of less importance.