Even in an ideal situation where a mutation has a "pure" benefit to the group, with no other drawbacks to the mutated gene, it will not be selected for. Selection pressure is a scarce resource. It protects existing adaptations from the ever-present entropic decay of mutation. Genes that are not selected for, relative to the other genes in the population, will reliably fall apart due to lack of maintenance and their effects will disappear. A gene must continually justify its survival by outperforming competing mutant alleles. It's just not true that group selection is merely weak, or a secondary effect occurring only when genetic selection allows it. The aggregate contribution of group selection to a gene's survival is literally zero.
I'm not arguing that music isn't adaptive. It's plausible that its adaptive value is to display to potential mates how well one's brain processes rhythm, which is useful for any number of physical activities as Pinker points out in the interview. But that's different than being selected for because it makes the group more cohesive. It may even, in fact, make the group more cohesive, but this will have nothing to do with why it was selected. This would have to be a happenstance byproduct, and thus not a plausible explanation for the existence of a complex adaptation like music. It would no more explain music than having your flowers watered would explain why it's raining.
So in what way is cancer not an example of group selection? Cancer cells reproduce more quickly, so in a sense they are fitter. However, on a large scale, the scale of complete groups (group = human being), they perform poorly because they kill the group. In other words, if the universe consisted of one giant human being then the cancer cells would definitely win. However, because of the group structure and the selection for complete groups, this doesn't happen.
I agree that within a group, group selection doesn't matter (duh). However, if you look at the aggregate of groups, after some time the groups with the group benefiting genes still exist while the other groups have died out. The groups that still exist will spawn new groups, etc, the standard evolution tale.
Consider this extreme case:
We have several groups. At every time step the group has a chance of being annihilated. An individual in a group can either have gene A or gene B. If an individual of a group has gene A then the chance that the group as a whole is annihilated is decreased. If it has gene B then this chance stays the same. Each individual in the group has a chance of reproducing. Once the group is above a certain size, it splits into two groups.
Now I'm sure you agree that gene A will be selected for relative to gene B, even though gene A gives no benefit to the individual relative to the other individuals in its group.
In the aggregate of groups, a gene with a huge benefit to the group and a slight disadvantage to the individual can, in the aggregate of groups, outperform a gene with a slight benefit to the individual but no benefit to the group. Sure, the conditions have to be just right for this to happen, but it does happen (e.g. cancer cells). So you can't say just because "group selection is impossible" that group selection doesn't happen in populations of animals. You need to analyze the particular situation to determine this.
It is possible, in principle, that rituals are another example of this.
I'm not arguing that music isn't adaptive. It's plausible that its adaptive value is to display to potential mates how well one's brain processes rhythm, which is useful for any number of physical activities as Pinker points out in the interview. But that's different than being selected for because it makes the group more cohesive. It may even, in fact, make the group more cohesive, but this will have nothing to do with why it was selected. This would have to be a happenstance byproduct, and thus not a plausible explanation for the existence of a complex adaptation like music. It would no more explain music than having your flowers watered would explain why it's raining.