Directed versus random
You will learn
When one carefully chosen case beats a thousand random ones, and when it does not.
A directed test is one case you chose because you know it bites: the empty FIFO, the full one, the wrap. A random one is cheap volume. They are not rivals; they are a sequence. The widget is the knockout experiment turned into a tool: three performance bits were claimed, and one knockout bitstream per bit -- the same design with that bit tied off -- proves each one matters. That is directed testing's whole argument: one bit, one prediction, one bitstream, measured.