I often find myself in conversation with supporters of Modern Monetary Theory, and for a long time I was mystified by what they are saying. This is the MMT I meet in argument and online, not the academic literature, which is more careful and more divided than any single account allows.
Over time I have come to think of MMT as three layers: a small set of accounting identities, a set of mechanisms, and a set of normative principles. Some are uncontroversial. Others are either inconsistent with the rest of MMT or simply impractical. What is most evident is that the identities do not establish the mechanisms, and the mechanisms do not establish the principles. If anything the arrow runs the other way: the mechanisms flow from the principles, and the identities are recruited afterward to make the whole look derived.
Three of the identities are mainstream economics; nothing new there. MMT adds a fourth claim and presents it as though it too were an identity, but it is not. It is the first of the mechanisms, an institutional design proposal wearing accounting's clothes. Of the six mechanisms, four rest on a technically true core the mainstream accepts but add reasoning or consequences it does not accept; the other two are institutional proposals the mainstream rejects outright. None of the principles are mainstream at all.
MMT did get one thing right, and it is worth saying so at the outset. A government that issues its own free-floating currency is not a household. It cannot be forced into nominal default on its own-currency debt, and even mainstream central banks now accept as much. So this is not the tired complaint that printing money causes inflation. The disagreement is about everything the theory builds on top of that one sound insight. I will take the layers in the order MMT usually presents them, from the arithmetic up to the politics, because that is the order you meet them in argument.