is there data backing up this n+1 claim? without A/B testing it, it is hard to be convinced that it is definitely the cheaper strategy. in fact, it makes a lot of assumptions about how people will behave, whats to prevent a culture developing where someone always grab the last roll but is responsible for flagging down the next batch? thus severely increasing your costs. regardless tho it is definitely non-intuitive and clever and worth testing for the sake of academic curiosity.