I thought it was interesting that this construct is linked in the paper to a slight variant of System F, which tells me that (a) it's probably pretty rigorously

I thought it was interesting that this construct is linked in the paper to a slight variant of System F, which tells me that (a) it's probably pretty rigorously rationalized (b) it can be utilized in future formal work, so (c) future formal work doesn't have to do closure allocation as much.