I will now discuss a macroscopic example of biological irreducible complexity and show how it can have evolved; this ought to make some of the principles clearer.
Honeybee societies are irreducibly complex. Queens depend on workers for all their needs; food, shelter, protection, etc. Workers depend on queens to replenish their numbers. Not surprisingly, honeybees found new hives by a queen leaving an existing hive with a swarm of workers.
So how could such societies have been a result of evolution from some species of solitary bee?
Michael Behe does the equivalent of looking at solitary bees and honeybees and wondering what could possibly bridge the gap.
But there are some social bee species that provide plausible intermediate states.
Bumblebees are social, yet bumblebee queens are not completely dependent on their workers as honeybee queens are. Bumblebee queens overwinter in isolation and found new hives in the following spring, doing all the work of caring for the first brood of workers until they mature. Thus acting like a solitary bee.
And a bumblebee-like bee can become the ancestor of honeybees by this route: departing queens hoping to found hives could recruit some workers to assist them. These prove very helpful -- so helpful that queens lose the ability to have an independent existence, thus becoming dependent on their workers.
Something similar could well have happened to the molecular pathways that Behe discusses; what would be especially helpful is finding or plausibly reconstructing some counterpart of bumblebees.