Note that there's nothing wrong with the definition of "irreducible complexity" per se. The assertion of biologists today is that the collection of irreducibly complex features is empty, and so far they're winning the debate. The example of the bacterial flagellum, for example, has already been answered.
So it is entirely possible for a scientist to look high and low for "irreducible complexity", and yet be perfectly scientific in his endeavor. What makes folks like Behe unscientific is that they ignore evidence when it runs counter to their views.
Not exactly. The debate is still very much alive. The problem for opponents of IC is that, unlike earlier arguments, which were based on ignorance, modern arguments are based on extensive knowledge. Behe declares the flagellum irreducible because we fully understand its operation.
The main problem with the alternatives given to explain away IC systems is that they remove the bias from the evolution equation (directed chance) and you begin working with random probabilities again, and even Dawkins will admit that evolution utterly collapses when it's a strictly random process.
An earlier post nailed this. ID gives a foot in the door to the "God people" and so is attacked for that reason and that reason alone.