That's true, I just meant to express (over-simplistically) that he argued that the mechanism is at the gene-level, which is even farther away from group-selection mechanisms.
The sections about emergent group-level behavior are really fascinating too.
You can still get completely unselfish individual traits out of selfish gene selection (like the entire chapter on runt gene selection). It looks like group selection, Dawkins only argues that the mechanism is selfish genes.
Even he seems to have forgotten he wrote about that though.