Here's what he has to say about evo-devo.
Consider the steady stream of recent books by authors striving to define a new field called evolutionary developmental biology (e.g., Arthur 1997; Gerhart and Kirschner 1997; Davidson 2001, 2006; Carrol et al. 2001; West-Eberhard 2003; Carrol 2005a; Kirschner and Gerhart 2005). The plots of all these books are similar: first, it is claimed that observations from developmental biology demonstrate major inadequacies in current evolutionary theory, and then a new view of evolution that eliminates many of the central shortcomings of the field is promised. Developmental biologists are correct in pointing out that evolutionary theory has not yet specifically connected genotype to phenotype's in a molecular/cell biological sense. However, extraordinary claims call for extraordinary evidence, and none of these treatises provide any formal example of the fundamental inability of evolutionary theory to explain patterns of morphological diversity. Those who argue that microevolutionary theory has made no contributions to our understanding of the evolution of form may wish to consult the substantial body of quantitative genetics literature on multivariant evolution. Such work is by no means fully satisfactory, as it is couched in terms of statistics (variances and covariances) rather than the molecular features of individual genes, but a more precise evolutionary framework for linking genes and morphology not be possible until a critical mass of generalities on the matter has emerged at the molecular, cellular and developmental levels.
For the vast majority of biologists, evolution is nothing more than natural selection. This view reduces the study of evolution to the simple documentation of differences between species, proclamation of a belief in Darwin, and concoction of a superficially reasonable tale of adopting the divergence (...). A common stance in cellular and developmental biology is that the elucidation of differences in molecular genetic pathways between two species (usually very distant species) completes the evolutionary story. No need to dig any deeper—because natural selection surely produce the end products, the population genetic details do not matter. In individual cases, this type of informal thinking may do little harm, but in the long-run it undermines the very scientific basis of with the evolutionary biology.
There are two fundamental issues here. First, the notion that interspecific differences at the molecular level reveal the mechanism of evolution ignores the fundamental distinction between the outcome of evolution and the events that lead to such changes. For example, although most animal developmental biologists argue that it was shocking to discover that the development of all animals is based on modifications of the same sets of ancient genes, many evolutionary biologists regard this view with some surprise. It is, of course, easy to criticize based on 20/20 hindsight, but we have known for decades that all eukaryotes share most of the same genes for transcription, translation, replication, nutrient uptake, core metabolism, cytoskeletal structure, and so forth. Why would we expect anything different for development? Although knowing that HOX genes play a central role in the development of all animals provides insights into the genetic scaffold from which body plans are built, it does not advance our knowledge of the evolutionary process much beyond noting that all vertebrates share a heritage of calcified skeletons. It need not even tell us that such genes were involved in the initial stages of differentiation (Alonso and Wilkins 2005). A vast chasm of stepwise (and partially overlapping) changes may separate today's products of evolution, and understanding those steps is what distinguishes evolutionary biology from comparative biology.