Haven’t read the book but did just look through some of the blog. It looks like he favors something akin to ‘natural genetic engineering’.
One issue for now: The author seems to think that the ENCODE project, which examined the sequences of transcribed DNA, demonstrated that there is very little or no junk DNA. Some of the papers published from the project ascribed ‘function’ or ‘biological activity’ to any portion of the genome that was transcribed into RNA. However, that a portion of the genome is transcribed is no indication of its status as ‘junk’ or whether it is dispensable. Gratuitous transcription (transcriptional noise) has been know for some time. There is also a solid theoretical and evidential basis for the understanding that most of our genome is composed of junk sequences.
Here is a link to a Scientific American blog that summarizes some of the discussion and includes links to other sources like Moran (‘Sandwalk’), T. Ryan Gregory (‘Genomicron’), Ed Young ('Not Exactly Rocket Science), and John Timmer (Ars Technica article).
Added: A fun post by Michael Eisen, a Howard Hughes Medical Institute Investigator at UC Berkeley.