I think the article is wrong, he assumes 100% automation. I'm not sure the marginal gain from completely eliminating all serving jobs will be worth it for a lon

I think the article is wrong, he assumes 100% automation. I'm not sure the marginal gain from completely eliminating all serving jobs will be worth it for a long time given that there's probably quite a long tail of quite complex interactions when things go wrong (and hey, redundancy is important as well in case the kiosks go down!). Though obviously the number of cashiers is going to end up significantly lower than it was at the start of this decade.