Recently Venumadhav et al. proposed a new pipeline to analyze LIGO-Virgo Collaboration's O1-O2 data, and discovered eight new binary black hole (BBH) mergers, including one with a high effective spin, χeff. This discovery helps to clarify the origin of the observed BBHs and the dynamical capture versus field binaries debate. Using a tide-wind model that characterizes the late phases of binary evolution and captures the essence of field binary spin evolution, we show that the observed χeff distribution favors this model over capture. However, given the current limited sample size, capture scenarios (isotropic models) cannot be ruled out. Observations of roughly 100 merges will enable us to distinguish between the different formation scenarios. However, if as expected, both formation channels operate, it may be difficult to resolve their exact fraction.