A couple of years ago, I identified two main usecases for blockchains: objective money/finance and objective identity

However, over time, I've warmed to the idea that objective identity does not actually require strict global consensus. A less strict, loosely ordered global consensus may be adequate

Really, it's only money that requires strict consensus, or anything else of *temporal* economic value

While identity does have economic value, it is in most cases non-temporal in nature, and can hence be achieved better with loosely ordered alternatives to blockchains

I agree with that statement. Although, rather than just "money," I might expand the definition to "value," which is broader and where it’s harder to draw a line. Ultimately, we’re talking about private property. The rest, I believe, only needs to be interoperable, but we could move toward an architecture that resembles a database. I think L3s or LXs may ultimately function like application databases, with certain properties of transparency and interoperability.