If I understand this paper correctly, I think switching our exponential mechanism for this method in our quantiles mechanism would be strictly beneficial, and I suspect that it could lead to major improvements in the high-ε case and/or for extremum quantile values.