...which, for some of us, isn't an argument, as we have basically no choice in the matter. But I'm still interested to know whether this is actually useful input or whether it's more holier-than-thou posturing.
To me, this is missing the forest for the trees. It's easy to argue about language implementation semantics when, in practice, it means almost nothing to anyone as we're stuck with the idiosyncrasies of the platforms we work on, even when we choose the platform.
I guess it's good that someone is looking at these things, but I can't see the utility in it. Best case scenario, this becomes a paradigm shift in new languages (which doesn't help us or anyone else for years) or gets implemented in the next major revision of language X (which doesn't go mainstream for years).