Feels like yet another example of "essential complexity driven by too much churn in infrastructural code".

I wonder why no one needs to write this article about dtrace probes? Is it because they are less used? Less capable? More stable? Better engineered?

Probably all of the above, alas.