Thanks for the great question, I just skimmed the paper. My reaction: it is still a vague proposal, with hand-waving, ill-defined concepts, and not at all axiomatically 'clean'. For example, he never defines «probability.»
Weinberg and others agree with t'Hooft at least in how to pose the problem: derive the probabilities from the deterministic unitary evolution. There have been real physical models done and published with this end in view, and they tend to take a quantum statistical mechanical approach, so there is some point of contact with some of t'Hooft's attitudes. But the valuable work in this way, as I see it, is using Schroedinger's equation to analyse actual physical measuring devices, such as the important work by Balian and two others at
arXiv:cond-mat/0203460 « Curie-Weiss model of the quantum measurement process.»
: See http://arxiv.org/abs/quant-ph/0507017 for a much less realistic toy model, and my axiomatically clean treatment of it's implications for Hilbert's Sixth problem, the axiomatisation of physics, http://arxiv.org/abs/0705.2554 ,
and Prof. t'Hooft is not even attempting to do that. It seems strange to hope to analyse measurement without thinking of the physics of measuring devices, or solve an axiomatic difficulty about probability without giving it a physical definition. I leave aside rival approaches to the problem, such as the decoherence approach, which some physicists are interested in.
Now QM seems to me, and most physicists, correct physically: the measurement problem is merely an axiomatic problem. Most physicists don't believe there is any new physics to be discovered which is relevant to the issue of determinism or the measurement problem, nor do I. (There are important physicists who are an exception, e.g., I suppose, Penrose.) I believe that a careful axiomatic analysis would be interesting, most physicists do not. I do not see one in this paper.