Remember that this is assuming that the two actors in the system are of equal levels of power.

I don't see how it is. You have the leader and the population affected. The leader has two decisions: (1) help subset x at the expense of y or (2) do nothing to protect subset y at the expense of x. There's no need to measure power or anything. This is a simple case of 1 individual's decision affecting multiple people. With either decision (action or inaction) people are harmed and benefitted.

That isn't counterintuitive at all. It's actually something quite a lot of people think is the better option, with fewer people living better lives.

Did you quote the wrong passage here? Because I was referring to how many many lives barely worth living being the best option is counter intuitive, but you responded saying that many people would find the idea of few lives with a lot of happiness the better option. Could you clarify which option are you saying is intuitive?

Which has been done and was considered moral, in the past. (in reference to slavery)

But that clearly isn't the best way to maximize happiness. Just because people thought that slavery was the moral action doesn't actually make it the moral action (moral being measured against well being, that is).

In my opinion that is evidence for the idea that, the theory of utilitarianism is too weak, it requires exceptions in order to function. (in reference to adding rules)

The rules will be based on the original end of well being. These rules (or rule), whatever they are, should be rules that generally maximize well being in the long run. That way its still consistent with the original aim of well being.