Most predictions for binary compact object formation are normalized to the present-day Milky Way population. In this talk, I suggest the merger rate of black hole binaries could be exceptionally sensitive to the ill-constrained fraction of low-metallicity star formation that ever occurred on our past light cone. I discuss whether and how observations might distinguish binary evolution uncertainties from this strong trend, both in the near future with well-identified electromagnetic counterparts and in the more distant future via third-generation gravitational wave detectors.