PZ Myers has a post on the evolutionary origins of complexity: αEP: Complexity is not usually the product of selection
I find it frustrating that people talk about terms that they don't clearly define, and assume everyone agrees on. Especially about complexity, which is hard to define, and I know not everyone has the same idea of. I'll just quote my comment on Pharyngula:
But, you have not quantified complexity, let alone say when there was an increase in it in the hypothetical example you give. If you don’t do this, you can’t talk about the evolution of complexity; it becomes a guessing game what we are talking about, and there is no chance that everyone will think of complexity as the same thing.
On top of that, there seems to be no distinction made anywhere between ‘complexity’ and ‘complex traits’. They need not be the same thing. Without defining complexity here(!), I’ll say that complexity can indeed easily arise by neutral processes, whereas complex traits cannot (but does have neutral and random processes involved) – it requires selection. And with that you’re going to ask me for a definition of a trait, so here is one: A single measurable component of the phenotype that has a function.
The key word here is function, without which I don’t know of any that can evolve without selection. Not selection every step of the way, as random processes are required (at least that’s how it occurs in nature), but selection at some point. The moment the trait acquires function, it becomes selected for.
On the other hand, ‘genomic complexity’ may not describe the state of a trait, but rather is the idea that the genome has many components that are intricately connected – which can arise by neutral processes.