The equation approximates the area under the curve. Increase the number of sampling from 11 to 101 say will increase the precision of the calculation. I don’t have a reason why it is 11. Is that just a legacy decision for VOC? Does that matter in practice for comparison purpose or are there some specific technical reason? Do we need such precision while there are other noises that make such precision illusional? Does the metric give unfair comparison so far?

But for my final answer, unfortunately, it is I don’t know.