To me, this seems like a productivity question. The software industry has the misguided notion that productivity can't be measured and
until that changes, it's very easy to argue that mob programming is efficient - because it's impossible to disprove (i.e. it's 'not even wrong'). In fact, because we don't have a concrete definition of what 'efficient' means, it's easy to argue any kind of programming is efficient. Big teams, small teams, work alone, OOP, FP, MVC, Java, Ruby - take your pick. We can all argue the merits of any pet methodology, language etc. In the end, what we really need is a way to measure individual productivity against a set of criteria relevant to a project.
I think this issue should be discussed a lot more.