They provide the basis for potentially taking on new functions, and constructive neutral evolution.
Adaptive evolution doesn’t have to always or even mostly be complexity-generating. The complexity generation is actually mostly through adaptive “degeneration” of expanding numbers of genetic elements, such as gene duplications. That’s the whole point here. Behe looks at life, sees all this complexy and all these functions, and concludes that because most molecular evolution is “degenerative” of functions, this means evolution can not have created all these functions and all this complexity we see, it must have had outside help. That’s his point.
But is that actually true? Does that logically follow from the observation that most molecular evolution is “degenerative”? Or is it possible that we can get new functions and more complexity through a process that adaptively “breaks” or “degrades” many more genes than it “creates” or “enhances”?
There’s nothing logically problematic about that. So yes, we can. Like this:
Red rectangles highlight what is being duplicated and passed on.
This is “adaptive devolution” of increased complexity, and new functions, by mostly “degrading” and mostly “breaking” genes. Because these extra genes are costly to express, their death is adaptive, and so is the eventual deletion of them. And because the still functional ones accumulate deleterious mutations because these are more frequent than beneficial ones, their duplication is also some times adaptive(more expressed genes compensates for each individual gene being weaker). Eventually a previously dead gene locus (effectively having become non-coding DNA) evolves into a de novo protein coding gene. So one new function is evolved and enhanced, while all the rest degrades and breaks. The net result is more complexity and more functions than there was to begin with. And it happened almost exclusively through neutral and adaptive degeneration.