Ill split hairs with you on this one. Change without selection is useless, selection without change is meaningless. Evolution is the selection of changes.
Selection acts at each of these levels, although in varying degrees.
Selection never operates on a group, only on the individuals in that group.
I think that Dawkins is correct in stating the absolute Darwinian unit is the gene. All evolution can be considered as genes trying to reproduce.
But, to always use this language is like trying to do chemistry by arguing from quantum mechanical principles. It is often more convenient to talk of individual or group selection, but this is a semantical convenience only.
A change can be adaptively neutral, but this is different from uselessness.
Selection never operates on a group, only on the individuals in that group.
Selection acts on phenotype or, more generally, traits. These traits can be borne by species, organisms, cell lines, genes, or molecules. The gene is a handy bookkeeping unit for documenting changes. But the gene is not the only unit of change, and is, only rarely, a unit of selection. Most traits in an organism, for example, are emergent from an aggregate of genetic sequences, expressed in the context of the whole organism.
Gould makes a good case for the species as the Darwinian individual. I come from a developmental perspective, and see changes defined by physical and developmental constraints, long before they become available for Darwinian selection.