The Problem of Temporary Extrinsics
I haven't really checked the literature, but is there a general agreement on why the problem of temporary intrinsics is a problem of intrinsics and not a general problem about temporary properties? Certainly it is just as impossible for a thing both to have and to lack an extrinsic property as it is impossible for intrinsic properties. A while ago, I said that perhaps for temporary extrinsics, the problem is not really a problem because the relational answer is the obviously correct one: having extrinsic property F at time t clearly means being F-related to t. But in fact that doesn't sound obvious at all. Does being an uncle relate people to times? It seems not. It seems only to relate them to other people. If one intuits that being round is not a relation to a time, I don't see why one wouldn't similarly intuit that being an uncle is not a relation to a time.
In the other post, I suggested that a better solution for many temporary extrinsics is to analyze away the predication completely: "I am an uncle in 2007" means "In 2007, one of my siblings has a child". But if this is agreed to be the solution, why is it never applied to temporary intrinsics (what I've called "Lowe's fourth solution" in the other post)? Anyway, it can only be the general solution if there are no primitive temporary extrinsic properties. It now seems to me that different solutions have to be applied to different temporary extrinsic properties: some are properties of temporal parts, some are analyzable away, and perhaps some are relations to times.
You're right that there's a problem here. If one takes intuitions about the adicy of properties/predicates to have strong evidential value, then there's a problem about 'uncle'. I think for Lewis intrinsics are a bigger problem because they are connected to change, and perhaps to naturalness. But in my impression (more from talking to people than the literature) there isn't really a consensus on what the problem of temporary intrinsics is.