We present recent results concerning Bock's direct multiple shooting method in the context of deterministic global optimal control. The introduction of artificial intermediate start values lifts the optimization problem to a higher dimensional space, compared to direct single shooting. At first sight, this looks like a very bad idea, as spatial branching schemes need to branch on more variables. Yet this lifting may yield significant advantages concerning the number of nodes to be processed and overall computation time, in addition to well-known features such as possibly improved local convergence rates, an improved stability, use of a-priori information for initial values, and effective parallelization. We shed some light on the issue why the lifting may also be beneficial concerning the size of the branching tree and illustrate the potential by application to benchmark problems from the literature.