There is a nice review out in BioEssays on the non-adaptive evolution of genome complexity. It focuses on the work of Michael Lynch and in particular a recent paper proposing a general theory about the evolution of genome complexity grounded on population genetics.
Lynch argues that the eukaryotic genome structure originated due to the reduced selection pressures associated with the small effective population sizes of eukaryotic species. That is, most of the complex features of eukaryotic genomes are the result of stochastic search through neutral or slightly deleterious mutations that are not purged trough selection (due to the reduced selection pressures in small populations).
I still have a hard time getting my head around this theory. I should try to strengthen my population genetics background. Lynch does not say that adaptation does not play a role. If I get it right he is just arguing that there is a relaxed search in reduced populations. This would be conceptually the same as relaxing a fitness search algorithm such that it would not get stuck in local optima (see fitness landscape).
One possibility to study this theory further could be maybe to take a realistic model of a simple cell and to do evolution studies in silico. Something like this was proposed recently in a Biology Direct paper by Neyfakh and colleagues.
I know some bloggers out there that would have a more informed opinion on this. It would be nice if there was some way to informally request an opinion just by linking in a particular way for example :).