Although their appears to be some "acrimony" in reference to Lyle's site http://www.geniebusters.org , I think at the very least he raises some good questions which remain unanswered in the scientific/technological communities.
The software complexity issue is still one that has not been fully addressed. Even the extropian icon Vernor Vinge raises similar doubts in his recent interview in Salon. Just how complex can things get? Are there bottlenecks to intentional design, or must we grow them with some combination of genetic algorithms, and real-time learning? If we can grow a complex intelligence, will it even be remotely similar to our brain structure that we can converse with it? And since it would appear an equally complex AI would have the same problems as ourselves in understanding its own brain functions, why would they have any fundamental advantage over increasing their intelligence at a faster rate than us? Wouldn't an human equivalent complexity increasing its own intelligence give us the same capability to increase our own?
I have one comment to make where I think Lyle misses an important point. His comment that diamondoid materials will no more become a universal material anymore than plastic or styrofoam is incorrect. Although diomondoid materials may not replace 'all' materials, they will become much more unbiquitous than any material previous to it. Obviously, the two key differneces is nanotechnolgies unique ability to integrate diomondoid material into any other material to any degree possible. More importantly, it is not the base material itself that is asimportant as it is the molecular mesostrcutures that nanotechnology allows us to interweave througout any structure. Such diverse mesostructures would allow one diomondoid structure to be completle flexible and another to be completely rigid, even though they weigh and look at exactly the same at the macroscopic level. Since certain carbon-nitrogen chains are the srongest and nearly the lightest material yet known, integrating them as fully as possible into every nanobuilt structure would seem to be the highest priority on every nanomanufacturer's list.