Another way of looking at the issue:

Imagine an intellgent queen, with a range of different types of sterile workers. Since the queen is smart, she it isn't limited to using the same genome in each type of worker - she can just put in the genes that are useful. Does this diversity reduce the level of cooperation between the workers? Not really - the genes of each worker have but one way to immortality - help the queen to reproduce.

In other words, the hypothesis that the level of cooperation depends on the proportion of shared genes is only a convenient rule of thumb, and should not be taken as being a golden rule.