Bob Murphy has a post up debunking Modern Monetary Theory here. His reaction is essentially the right one: it's bad economics to confuse accounting identities with behavioral laws. This goes for the MMTers G-T=S-I as much as it goes for the old MV=PQ. These identities can be extremely useful for pinning down the impact of a behavioral law, but you have to be careful.
Are Austrians guilty of this? Some are. Some point to a chart of monetary aggregates, reference something like MV=PQ, and think they've done economics. Many think Keynesians are guilty of this - they're convinced Keynesians think "a recession happens because people spend less so the government should spend more to make up for it". It's true Keynesians do think recessions happen because people spend less and the solution is for govenment to spend more - but there are a lot of behavioral claims behind that that aren't usually mentioned by critics. Similarly, "crude Austrians" ignore their own claims about the capital structure which fill out simplistic accounting identities.
Anyway - the point is, economics is not accounting. Good for Bob Murphy for pointing that out. If Bob or I have an incomplete understanding of MMT, I'd appreciate elaboration. I haven't put all that much time into learning about it. It would probably be most useful to elaborate by identifying what sort of behavioral processes you think underly each of the components of that accounting identity, because as Bob ably points out you can get everything from Krugman and me to the Tea Party out of G-T=S-I, depending on what assumptions you bring to the model.