I think the problem with the MVP is that it's hard to identify before you build it.
Or rather: it's easier to bring a full-scale app down to an MVP, than to build a successful MVP from the ground up.
Some features just don't pan out to be what you initially hoped they would, and others get used in ways you'd never envisioned.
Anyway, this is mostly dependent on what type of solution you're solving. Some things just are simple enough to build an MVP.
I agree—the scale of the project informs how big/small an MVP can be.
My biggest concern with MVPs is that they serve as an easy “out.” I know I’ve personally abandoned projects too early, because the uptake wasn’t there.
In most cases, I’ve later regretted having done so.