We present an analysis of how pointing performance in in-air un-instrumented pointing can be improved, towards throughput equal to the mouse. Pointing using a chopstick is found to achieve the highest average throughput, with 3.89 bps. This is a substantial improvement over using the finger to point at the screen. Two potential reasons for the throughput gap between chopstick and finger operation were explored: the natural curvature of human fingers and tracking issues that occurs when fingers bend toward the device. Yet, neither one of these factors seems to significantly affect throughput. Thus other, yet unexplored factors must be the cause. Lastly, the effect of unreliable click detection was also explored, as this also affects un-instrumented performance, and was found to have a linear effect.