Flight characteristics are reported to be the same in the normal flight envelope except at high angles of attack where the new engine nacelle actually produces lift which translates into a further pitch up - but I have no idea how aggressive this is. But it's enough that this is why MCAS exists.

I'm certainly concerned whether MCAS is both working as intended, and also failing safe. But my more immediate concern is like yours: when MCAS is effectively disabled in the normal course of troubleshooting runway trim by setting stab trim to cutoff, now you have a plane that has different stall behavior than you're type rated for! Flip those switches, now you need a different type rating! Of course pilots can learn different stall behaviors and avoidance for a new type, we don't need abstraction to do that for us when we know about it and have trained for it. Stall avoidance is fundamental make+model knowledge, but it can be non-obvious and making it obvious and deliberate is what the type rating is about.

So yeah we might actually end up in the very curious case where either Boeing, or FAA or NTSB are all: this is going to require a type rating afterall. The very thing the airlines in particular want to avoid.

It seems to me the in-cockpit "aoa disagree" option needs to become standard by AD. MCAS only takes input from one of the two alpha vanes, so if it gets bogus data it has no backup source. Meanwhile the pilots have no indication the two vanes disagree unless they (apparently) bought that indication feature. Which I think is pretty fucked up if that turns out to be true and relevant as to either cause or solution in all of this.