The magnetically driven superconductor-insulator transition in amorphous thin films ͑e.g., InO and Ta͒ exhibits several mysterious phenomena, such as a putative metallic phase and a huge magnetoresistance peak. Unfortunately, several conflicting categories of theories, particularly quantum-vortex condensation, and normal region percolation, explain key observations equally well. We present a experimental setup, an amorphous thin-film bilayer, where a drag resistance measurement would clarify the role quantum vortices play in the transition, and hence decisively point to the correct picture. We provide a thorough analysis of the device, which shows that the vortex paradigm gives rise to a drag with an opposite sign and orders of magnitude larger than the drag measured if competing paradigms apply. DOI: 10.1103/PhysRevB.80.180503 PACS number͑s͒: 74.78.Db, 73.43.Nq, 74.25.Fy, 74.78.Fk The superconducting state and the metallic Fermi-liquid form the very basis of our understanding of correlated electron systems. Nevertheless, the transition between these two phases in disordered films is shrouded in mystery. Experiments probing this transition in amorphous thin films such as Ta, MoGe, InO, and TiN used a perpendicular magnetic field and disorder ͑tuned through film thickness͒ to destroy superconductivity. But instead of a superconductor-metal transition, they observed in many cases a superconductor-insulator transition ͑SIT͒. 1 The "dirty boson" model 2 propounded the notion that the insulator is the mark of vortex condensation, and that the SIT occurs at a universal critical resistance, R ᮀ = h / 4e 2 . More recent experiments, however, showed the critical resistance to be nonuniversal. 3 Furthermore, in many field tuned experiments, a surprising metallic phase intervenes between the superconductor and insulator, 4-6 with a temperature-independent resistance below T ϳ 50 mK, and ͑at least in Ta films͒ a distinct nonlinear I-V characteristics. 7 Quite generically, 5,6,8 these films exhibit a peak in the magnetoresistance ͑MR͒ curve ͑particularly strong in InO and TiN͒ as in Fig. 1͑a͒.Two competing categories of theories may account for these phenomena. On one hand, within the quantum vortex pictures, 2,9,10 the insulating phase implies vortex condensation, the intervening metallic phase is described as uncondensed vortex liquid ͑e.g., vortex Fermi liquid͒, and the high field nonmonotonic MR indicates the appearance of a finite electronic density of states ͑DOS͒ at the Fermi level. On the other hand, the percolation paradigm 11,12 describes the films as consisting of superconducting ͑SC͒ and normal puddles; at the MR peak SC puddles exhibit a Coulomb blockade, and the percolating normal regions consist of narrow conduction channels. Yet a third theory tries to account for the low field SC-metal transition using a phase glass model 13 ͑see, however, Ref. 14 which argues against these results͒ but does not address the full MR curve. Qualitatively, both paradigms above are consistent with MR observations, and recent...