If you're open to a bit different and perhaps thought provoking perspective let me say that for your hypothesis to be true we would need to assume that the dist

If you're open to a bit different and perhaps thought provoking perspective let me say that for your hypothesis to be true we would need to assume that the distribution of memory-safety bugs is uniform across the codebase. From my experience this is not true, it usually follows uni-modal or at most bi-modal distribution, so it follows that the surface is not much smaller.