Ok so what is the point of neurotically obsessing about intelligence? Would it not be more fruitful to emphasize computation?

Do note that the orthogonality thesis is a hypothesis, not something we have demonstrated. Weak versions of it (e.g. it is possible to have an intelligent agent with arbitrary goals) are more likely to be true than stronger versions (e.g. intelligent agents we build will have goals uniformly selected from the space of all possible goals).