The core significance "rational interests" is that they are consistent. For example, if you like A more than B and B more than C, you should like A more than C.

The core significance "rational interests" is that they are consistent. For example, if you like A more than B and B more than C, you should like A more than C. The important part is the internal consistency, not the actual preferences. You could aim at nothing more than helping your fellow man and behave perfectly rationally, as long as you're consistent about it. It's not a perfect model because people's preferences are not necessarily totally ordered, and people are not always consistent, and preferences change in interesting but sometimes inconsistent (and often predictable) ways. But it's pretty good. I think it's a great way to think about things because it makes no judgements about what your actual goals are. After all, people have different preferences and interests, and that's fine. More importantly, this also means you could reuse the same ideas perfectly happily for corporations, organizations, societies or event AIs and still draw interesting conclusions. It also lets us create systems that can successfully cater to agents with wildly different interests and get them all to cooperate willingly—that's one of the most important features of our market system and capitalism.