We can solve the issue another way, by finding methods to reduce the fees (smartfees etc).
How low do you think they would go? Wasn't there a calculation somewhere about the potential block orphan probability vs extra bandwidth. That set the current (or one of the previous) fees.
But for a hub/spoke design, you could start with a single hop implementation and find ways to extend it to multi-hub with pathfinding later.
True.
I guess with mirco-payments, risk is low anyway. You send a payment over 2 hops and if you are ripped off, you don't use that hub anymore.
The hubs would similarly not forward payments via other nodes that have ripped them off.
Some kind of proof of receipt would be needed though.