While I can see this as a valid experiment in the digital world where each additional copy doesn't cost you anything, it doesn't make much sense when each item has a cost to produce.
I can see why it might not work because each item costs something to produce, but that doesn't make it an invalid or uninteresting experiment.
My expectation would be that the experiment will fall with the current model (put some money in a jar) - I would guess they would do better with something less anonymous, with more room for social pressure.
On the other hand, four is shockingly cheap when bought in bulk, and service workers are also not particularly expensive, so it's quite possibly they'll be 'facebook profitable' and if you've got a large corp willing to cover your CapEx, that's good enough.
An interesting side effect of money in the jar though is that people are socially pressured not to take change from the jar, meaning they may pay say $5 instead of the $3 they normally would because they had the note.
My expectation would be that the experiment will fall with the current model (put some money in a jar) - I would guess they would do better with something less anonymous, with more room for social pressure.
On the other hand, four is shockingly cheap when bought in bulk, and service workers are also not particularly expensive, so it's quite possibly they'll be 'facebook profitable' and if you've got a large corp willing to cover your CapEx, that's good enough.