Can someone with an economics background tell me whether this is a sound argument? It looks like it could be reasonable to me, but it also looks a lot like the examples of deadweight losses in Bastiat's "That Which is Seen, and That Which is Not Seen".
I would guess that we should count the $3T transferred as a cost on one side of the book and as a benefit on the other side, and from an accounting standpoint not care much either way. I'd also guess we should somehow try to compare the relative value of the service on each side of the transaction to gauge somw "surplus" value. People get food they value at $3.1T, and they pay $3T, and the workers would have done the job for $2.9T, so society wins by $200B in aggregate. Or something. But maybe that's nonsense?
I don't know if that's reasonable, but I am fairly convinced that we shouldn't dig $3T worth of ditches just to fill them in. That is deadweight loss.
At the very least I think we can assume that restaurants are not a deadweight loss because they're profitable, though. People often choose to pay that premium over cooking for themselves, and we should trust that those choices reflect personal relative value well.