The salmon example is a really clever way of separating some of the issues.

I think the salmon example is seriously misleading, and in a way that shows a very common pattern of fallacies in consequentialist reasoning. It presents a thought experiment that is contrived to be free of any game-theoretic concerns, and then this example is used as a rhetorical sleight of hand by positing a superficial analogy with a real-life example, in which the game-theoretic concerns are of supreme importance.

Subsequently, these concerns are dismissed with another misleading observation, namely that people rarely fake offense. Well, yes, but the whole point is that people's sincerely felt emotions are very much directed by their brains' game-theoretic assessment of the situation, which may well indicate that a seemingly irrational extreme emotional response is in fact quite rational given the circumstances. Those who ignore this point should read up on their Schelling.