Recently Venumadhav et al. (2019) proposed a new pipeline to analyze LIGO-Virgo's O1-O2 data and discovered eight new binary black hole (BBH) mergers, including a high effective spin, chi_{eff}, one. This discovery sheds new light on the origin of the observed BBHs and the dynamical capture vs. field binaries debate. Using a new statistical approach, we show that, while isotropic models are not ruled out, the observed chi_{eff} distribution favors field binaries, whose chi_{eff} is determined by tidal forces and wind losses, over capture. However, given the current limited sample size, capture scenarios (isotropic models) cannot be ruled out. Observations of roughly a hundred 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.