The direct collapse model for the formation of massive black holes has gained increased support as it provides a natural explanation for the appearance of bright quasars already less than a billion years from the Big Bang. In this paper we review a recent scenario for direct collapse that relies on multi-scale gas inflows initiated by the major merger of massive gas-rich galaxies at z > 6, where gas has already achieved solar composition. Hydrodynamical simulations undertaken to explore our scenario show that supermassive, gravitationally bound compact gaseous disks weighing a billion solar masses, only a few pc in size, form in the nuclei of merger remnants in less than 105 yr. These could later produce a supermassive protostar or supermassive star at their center via various mechanisms. Moreover, we present a new analytical model, based on angular momentum transport in mass-loaded gravitoturbulent disks. This naturally predicts that a nuclear disk accreting at rates exceeding yr−1, as seen in the simulations, is stable against fragmentation irrespective of its metallicity. This is at variance with conventional direct collapse scenarios, which require the suppression of gas cooling in metal-free protogalaxies for gas collapse to take place. Such high accretion rates reflect the high free-fall velocities in massive halos appearing only at z < 10, and occur naturally as a result of the efficient angular momentum loss provided by the merger dynamics. We discuss the implications of our scenario on the observed population of high-z quasars and on its evolution to lower redshifts using a semi-analytical galaxy formation model. Finally, we consider the intriguing possibility that the secondary gas inflows in the unstable disks might drive gas to collapse into a supermassive black hole directly via the General Relativistic radial instability. Such dark collapse route could generate gravitational wave emission detectable via the future Laser Interferometer Space Antenna (LISA).