Yes, the window analogy is more or less accurate. Let's go with the simplistic version of collapse just for the idea. A measurement P collapses the wave function randomly into an eigenstate of P. Then if a different measurement Q is made the wave function will randomly collapse into an eigenstate of Q. So for example if I measure position, the wave function will collapse into a narrow peak. Now if I measure momentum, the wave function will collapse into a spread out wave. If I alternate between position and momentum measurements, the wave function will keep jumping between being peaked and spread out. So in Zurek's analogy, each tourist is making a different measurement and so collapsing into an eigenstate of the respective measurements, so reality will be all jumpy.
In addition to Zurek's approach, other lines to explaining the conditions under which repeated or continuous measurements give classical results are:
http://arxiv.org/abs/1305.2517
http://arxiv.org/abs/1407.8090
I think in those papers, one does get a stochastic differential equation describing a jumpy reality. But as long as the jumps are "small", one will have recovered classical trajectories.