Behe makes a lot of the impossibility of a species getting from a local peak to the global peak, because of valleys between the peaks., if natural selection was the only force driving evolution.
Gavrilets (2004), however, only shows these images at the very beginning of his book, in order to prove them wrong. As fitness landscapes are multidimensional (not 3 dimensional), there are many connections between peaks. As a mental picture imagine bridges between the peaks bridging valleys. Therefore Gavrilets concludes that fitness landscapes are very different from ordinary landscapes and that populations form networks within them, where much of the variation is neutral in terms of fitness. Nevertheless Behe presents the rugged landscape as if it was the state of art in research and makes a lot of fuss about the impossibility of a population crossing valleys in the landscape.
How is that to be judged? In my eyes, it is either gross negligence or dishonesty. If Behe took the images from Gavrilets without taking even a superficial look at the context and what Gavrilets was showing, it would be gross negligence. If, however, he knew what Gavrilets was about, it would be dishonest.
Given that on page 114, Behe falsely credits Fisher (not Wright) with conceiving the idea of a fitness landscape, I conclude that he is only negligent or just ignorant. Otherwise he would have known that Wright already proposed drift in small sub-populations as a means of crossing valleys that requires no ID.
